TechPrufer
Get started
Cloudflare
Cloudflare logoCloudflare · Feature trail

Global Network Expansion

313 posts · Jan 2011 – Jul 2026

What this trail covers

Cloudflare's global network has expanded to include new data centers and increased server capacity across existing locations, enhancing performance and reliability for users worldwide. This expansion aims to reduce latency, improve bandwidth availability, and provide regional redundancy by deploying infrastructure closer to end-users. The addition of new locations and increased capacity contributes to handling growing traffic demands. The network now exceeds 500 Tbps of external capacity, with a focus on building out regional backbones to reduce traffic tromboning and support data localization.

All Cloudflare trails
  1. BGP ORIGIN attribute manipulation and its impact on the InternetThis post details an investigation into the manipulation of the BGP ORIGIN attribute, revealing that approximately 70% of observed paths have a different ORIGIN value than what was set by the originating Autonomous System. The analysis used BGPKIT and MRT dumps from RIPE RIS and RouteViews, along with internal BMP data, to identify ASes that rewrite the ORIGIN attribute, primarily to IGP, to influence route selection and divert traffic. The study found significant manipulation in both IPv4 and IPv6 routing, with nearly 10% of direct peers rewriting the ORIGIN to IGP for IPv4. The findings highlight the revenue-driven nature of BGP route selection and the widespread deviation from RFC standards.
  2. Enforcing the First AS in BGP AS PATHsThis post details an investigation into route hijacks facilitated by forged BGP AS PATHs. It analyzes specific hijack examples, explains the vulnerabilities in BGP related to First AS checking, and proposes the enforcement of First AS matching the peer AS as a critical security measure. The post also describes an experiment where Cloudflare intentionally violated the First AS rule to measure its acceptance on the internet.
  3. Iran's Internet is partially restored, Cloudflare Radar data showsThis post leverages Cloudflare Radar data to analyze and confirm the partial restoration of internet access in Iran following a prolonged national shutdown. It details the timeline of two distinct shutdowns, analyzes traffic and DNS query increases as indicators of restoration, and highlights regional and network-specific traffic patterns. The post also notes the continued impact on IPv6 connectivity while IPv4 remains relatively stable, suggesting specific technical means were used for the shutdown.
  4. Shutdowns, power outages, and conflict: a review of Q1 2026 Internet disruptionsThis post details observed internet disruptions in Q1 2026, categorizing them by cause: government-directed shutdowns (Uganda, Iran, Republic of Congo), military action (Ukraine, AWS Middle East), power outages (Cuba), severe weather (Portugal), cable damage (Republic of Congo), and technical problems (Verizon Wireless, Guinea, UK). It provides specific data points on traffic drops, IP address space changes, and the duration of these disruptions, illustrating the impact on global connectivity and Cloudflare's network monitoring capabilities.
  5. Agents Week: network performance updateThis post details Cloudflare's improvement in network performance, becoming the fastest provider in 60% of the top 1,000 networks globally by December 2025, up from 40% in early 2025. The improvements are attributed to both hardware expansion (new PoPs in Constantine, Algeria; Malang, Indonesia; and Wroclaw, Poland) and software optimizations. Software improvements include leveraging protocols like HTTP/3, refining congestion window management, and optimizing the efficiency of connection handling, SSL/TLS termination, and traffic management software. The post also elaborates on the methodology for measuring network performance using connection time and the trimean of Real User Measurements (RUM) data.
  7. 500 Tbps of capacity: 16 years of scaling our global networkThis post details Cloudflare's network growth to 500 Tbps of external capacity across 330+ cities. It describes the evolution from early transit and peering relationships to a distributed security layer. Key technical contributions include: the architecture of the DDoS mitigation system using eBPF (_xdpd, _l4drop, _dosd) and Quicksilver for global rule propagation; the integration of edge computing (Workers, Containers) on the same infrastructure that handles DDoS; and the ongoing adoption and promotion of forward-looking protocols like RPKI and ASPA for BGP security. It also discusses the engineering challenges of distinguishing AI traffic from human traffic using TLS fingerprinting and behavioral analysis.
  8. ASPA: making Internet routing more secureThis post introduces ASPA (Autonomous System Provider Authorization) as a new cryptographic standard to enhance BGP security by validating the entire path of network traffic, building upon RPKI. Cloudflare Radar has launched a new feature to monitor ASPA deployment trends. The post details how ASPA works, its validation process, its effectiveness against forged-origin hijacks (with noted limitations), and the ease of creating ASPA objects within RIRs like RIPE and ARIN. This directly contributes to the network's security and reliability by addressing a critical challenge in internet routing.
  9. Cloudflare outage on February 20, 2026This post details a significant outage on February 20, 2026, caused by a misconfiguration in the Bring Your Own IP (BYOIP) service. A change intended to automate the removal of prefixes from the BYOIP pipeline contained a bug that led to the unintentional withdrawal of approximately 1,100 customer prefixes from the Cloudflare network. This resulted in services and applications being unreachable for affected customers. The post outlines the impact on various services, the root cause related to the Addressing API and a faulty cleanup sub-task, the limitations of staging environments and testing in catching this specific bug, and the challenges in immediate recovery due to different states of impact. It also highlights the ongoing 'Code Orange: Fail Small' initiative aimed at improving deployment safety and rollback capabilities.
  10. Cable cuts, storms, and DNS: a look at Internet disruptions in Q4 2025This post details observed internet disruptions in Q4 2025, categorizing them by cause: government-directed shutdowns, cable cuts (Digicel Haiti, Cybernet/StormFiber Pakistan, Camtel/MTN/Orange Cameroon, Claro Dominicana), power outages (Dominican Republic, Kenya), military action (Odesa, Ukraine), and technical problems. It quantifies the impact of these events on traffic and announced IP address space, leveraging Cloudflare Radar data and external reports to provide a comprehensive overview of global internet stability during the quarter.
  11. Route leak incident on January 22, 2026This post details a specific incident of a BGP route leak caused by an automated routing policy configuration error in the Miami data center. It explains the technical cause of the leak (a permissive policy matching 'internal' route types and advertising them externally), the impact (25 minutes of congestion, elevated loss, higher latency, discarded traffic), and the timeline of detection and resolution. It also provides technical details on BGP route leak definitions and analysis using tools like monocle and RIPE BGPlay.
  12. A closer look at a BGP anomaly in VenezuelaThis post analyzes a BGP route leak anomaly observed in Venezuela, attributing it to potential insufficient routing export and import policies by the ISP CANTV (AS8048). It explains BGP and route leaks, details the specific leak involving AS8048 redistributing routes from AS6762 to AS52320, and discusses the relationship between AS8048 and the originating network AS21980. The post contrasts the observed leak with potential malicious intent, suggesting configuration errors or network issues as more likely causes, and highlights the role of tools like Cloudflare Radar and BGPKIT's monocle in analyzing such events. It also touches upon RFC9234 and the Only-to-Customer (OTC) attribute as potential solutions for preventing route leaks.
  13. BGP zombies and excessive path huntingThis post details the formation and impact of BGP zombies and excessive path hunting, explaining how route withdrawals can lead to temporary routing loops and reachability issues. It describes the underlying causes, including router software bugs, processing slowness, and the Minimum Route Advertisement Interval (MRAI). The post also illustrates how Cloudflare's BYOIP on-demand advertisement feature can increase the frequency of these events and presents observed instances of BGP zombies within upstream partners and Cloudflare's own network. It highlights the longer convergence times observed for IPv4 compared to IPv6.
  14. A framework for measuring Internet resilienceThis post introduces a data-driven framework for measuring Internet resilience, defining it as the measurable capability of a national or regional network ecosystem to maintain diverse and secure routing paths and rapidly restore connectivity. It highlights the importance of local decisions having global impact, the role of route hygiene (RPKI/ROV) in network health, and the challenges of measuring resilience due to data access limitations. The post details metrics related to IXPs and colocation facilities, emphasizing their role in local and global interconnection, and the need for diverse peering participation.
  15. Internet measurement, resilience, and transparency: blog takeover from Cloudflare Research and friendsThis post announces a week-long series of blog posts from Cloudflare Research and friends, focusing on internet measurement, resilience, and transparency. It highlights upcoming content on foundational internet measurement concepts, Cloudflare Radar's evolution, and the workings of their internet speed test. It also details plans to introduce a framework for internet resilience, discuss post-quantum cryptography advancements, and explore network characteristics at Cloudflare scale. The post emphasizes the role of cryptographic protocols in securing interactions with AI agents and crawlers, building upon Web Bot Auth and proposing new protocols for honest users.
  16. Monitoring AS-SETs and why they matterThis post introduces Cloudflare Radar's new feature for monitoring AS-SETs, which are crucial for BGP routing and building route filters. It explains how AS-SETs are used to group related networks, their current trust-based limitations, and the upcoming ASPA standard. The post details how accurate AS-SETs prevent route leaks, highlights AS-SET misuse, and showcases the new Radar feature that provides transparent, public listings of AS-SET information, including inferred ASN, IRR sources, member counts, and AS cone sizes, to help network operators monitor and manage their routing configurations.
  17. Introducing new regional Internet traffic and Certificate Transparency insights on Cloudflare RadarThis post introduces significant enhancements to Cloudflare Radar, extending its visibility into Internet traffic and security. It adds regional traffic insights, allowing for more localized analysis of trends, and integrates detailed Certificate Transparency (CT) data, building upon Cloudflare's existing work in this area. These additions provide deeper, more granular visibility into the health and security of the Internet at both sub-national and cryptographic levels.
  18. Cloudflare incident on August 21, 2025This post details a network congestion incident caused by a single customer's traffic surge overwhelming links between Cloudflare and AWS us-east-1. It explains the technical details of the congestion, including the role of BGP prefix withdrawals by AWS, saturated peering connections, and overloaded Data Center Interconnect (DCI) links. It also outlines short-term remediations like traffic deprioritization, expedited DCI upgrades, and improved BGP traffic engineering with AWS, as well as a long-term solution involving a per-customer traffic management system.
  19. Russian Internet users are unable to access the open InternetThis post details how Russian ISPs are throttling internet traffic, impacting Cloudflare users within Russia. The throttling limits web asset downloads to 16 KB, rendering most navigation impossible. Cloudflare is unable to restore access due to the throttling occurring within Russian ISPs and outside of Cloudflare's control. Data analysis, including traffic trends, Network Error Logging (NEL), and packet loss metrics, corroborates the throttling.
  20. Bringing connections into view: real-time BGP route visibility on Cloudflare RadarThis post introduces a new real-time BGP route lookup service on Cloudflare Radar, accessible via both a web interface and an API. This service provides near-instantaneous visualizations of BGP routes for IP prefixes, including Sankey diagrams and detailed origin tables with RPKI validation status. The accompanying API allows developers to programmatically access this data for automated tools, enhancing capabilities for BGP leak and hijack detection. The feature leverages data from RIPE RIS Live and RouteViews to reduce data delay from minutes to seconds, significantly improving the ability to monitor and debug internet routing.
  21. How the April 28, 2025, power outage in Portugal and Spain impacted Internet traffic and connectivityThis post details the impact of a major power outage in Portugal and Spain on internet traffic, network quality, and routing. It provides specific metrics and graphs showing traffic drops, changes in network quality (download speeds, latency), and routing impacts (announced IP address space). The analysis highlights how Cloudflare's global network and customer base provide unique insights into internet resilience during large-scale infrastructure failures.
  22. The fall and rise of TikTok (traffic)This post analyzes the impact of the US ban on TikTok on DNS traffic, detailing the timeline of service disruption and restoration. It highlights how Cloudflare's 1.1.1.1 DNS resolver and autonomous system data provided real-time insights into the traffic drop and subsequent recovery. The post also details the surge in DNS traffic for TikTok alternatives like RedNote (Xiaohongshu) in the US and globally, showcasing Cloudflare's ability to track shifts in internet usage patterns due to geopolitical events and regulatory actions.
  23. Multi-Path TCP: Revolutionizing connectivity, one path at a timeThis post introduces Multi-Path TCP (MPTCP) as a significant advancement in internet connectivity, building upon Cloudflare's existing efforts in network expansion and protocol optimization. It details how MPTCP leverages multiple network interfaces (e.g., Wi-Fi and cellular) to create 'subflows' for a single TCP connection, enhancing aggregation and mobility. The post explores MPTCP implementations on Linux and iOS/macOS, discusses configuration for server and client roles, and highlights the challenges and ongoing development in path management and scheduling. It also touches upon the potential for MPTCP to improve reliability for mobile clients and in data center scenarios, aligning with Cloudflare's goals of a faster, more reliable internet.
  24. Resilient Internet connectivity in Europe mitigates impact from multiple cable cutsThis post details how Cloudflare's global network infrastructure, particularly in Europe, demonstrated resilience against multiple submarine cable cuts. It analyzes traffic volume indicators, internet quality metrics (bandwidth and latency), and BGP announcements for Lithuania, Sweden, Finland, and Germany. The analysis shows minimal to no observable impact on traffic and quality, attributing this to the significant redundancy and alternative routing paths available through multiple submarine cables and terrestrial fiber networks in the region. The post uses Cloudflare Radar data to illustrate these findings and references existing submarine cable maps and network provider maps to support the claims of infrastructure resilience.
  25. Cloudflare’s perspective of the October 30, 2024, OVHcloud outageThis post details Cloudflare's observation and analysis of the OVHcloud outage on October 30, 2024. It includes traffic graphs showing the impact on Cloudflare's origin traffic, analysis of BGP routing shifts, and the discovery of evidence pointing to a BGP route leak from Worldstream (AS49981) impacting OVHcloud. It also highlights Cloudflare's own protective measures, such as max prefix-limits, which prevented the route leak from affecting Cloudflare's network.
  26. Impact of Verizon's September 30 outage on Internet trafficThis post details how Cloudflare's global network visibility and analytics capabilities were used to observe and report on the impact of a major Verizon mobile network outage. By analyzing HTTP request volume from Verizon's autonomous system (AS6167) and specific impacted cities, Cloudflare provided real-time insights into the decline and subsequent restoration of internet traffic, highlighting the network's ability to monitor and understand large-scale service disruptions.
  27. Removing uncertainty through "what-if" capacity planningThis post introduces "Scenario Planner," a new system for capacity planning that allows internal teams to simulate "what-if" scenarios related to demand and supply changes across Cloudflare's global data centers. It details the system's inputs (demand and supply changes), calculations (conversion to CPU Time and distribution across regions), and outputs (heatmaps and "Expected Failovers" views). The post also elaborates on the importance of CPU Time as a unit of measurement for capacity planning and provides examples of demand and supply "what-if" questions and their corresponding inputs and calculations within Scenario Planner.
  28. Cloudflare incident on September 17, 2024This post details an incident on September 17, 2024, where Cloudflare inadvertently stopped announcing fifteen IPv4 prefixes due to an internal software error during a routine renumbering exercise. This led to a temporary outage for some Business plan websites. The post explains the background of the Addressing team's work on IP address management and renumbering, the specific software bug that caused the failure (related to legacy SSL configurations and incomplete website renumbering), the incident timeline, and the immediate fix (rollback). It also outlines preventative measures including enhanced testing for outstanding assignments, improved verification processes for prefix withdrawals, and a review of legacy system functions.
  29. Making progress on routing security: the new White House roadmapThis post details the White House roadmap for enhancing internet routing security, focusing on the importance of BGP security and the adoption of RPKI. It explains how BGP vulnerabilities can lead to hijacks and leaks, and how RPKI, through Route Origin Authorization (ROA) records and Route Origin Validation (ROV), provides a cryptographically secure mechanism to validate IP prefix origins. The post also discusses past incidents of BGP hijacks and leaks, emphasizing the need for widespread RPKI adoption to mitigate these risks and improve overall internet routing security.
  30. The backbone behind Cloudflare’s Connectivity CloudThis post details Cloudflare's significant expansion of its global backbone network since 2021, increasing capacity by over 500% to over 330 cities in 120 countries. It explains the strategic importance of this backbone for traffic routing control, cost-effectiveness, and enabling products like Tiered Caching and Magic Transit. The post highlights the use of dark fiber and DWDM for inter-data center connectivity, and the benefits of regional backbone investments in reducing traffic tromboning and supporting data localization. It also quantifies latency improvements, showing up to 22% faster content delivery in Johannesburg due to backbone utilization.
  31. A recent spate of Internet disruptionsThis post details recent internet disruptions observed by Cloudflare Radar in Bangladesh, Syria, and Pakistan during July 2024. It highlights government-ordered shutdowns in Bangladesh due to protests, scheduled exam-related shutdowns in Syria, and a large-scale disruption in Pakistan affecting major ISPs. The post provides specific data on traffic drops, IP address space changes, and the timeline of service restoration for each region, underscoring Cloudflare's role in monitoring and reporting on global internet stability.
  32. Cloudflare 1.1.1.1 incident on June 27, 2024This post details a specific incident on June 27, 2024, where a BGP hijack of 1.1.1.1/32 by ELETRONET S.A. (AS267613) and a route leak of 1.1.1.0/24 by Nova Rede de Telecomunicações Ltda (AS262504) caused temporary unreachability and high latency for the 1.1.1.1 DNS resolver. It explains the technical details of BGP hijacks and route leaks, the incident timeline, and the impact on users, emphasizing Cloudflare's commitment to improving detection and advocating for RPKI adoption to prevent future occurrences.
  33. How the UEFA Euro 2024 football games are impacting local Internet trafficThis post analyzes the impact of UEFA Euro 2024 on local internet traffic, observing an average drop of 6% in European countries during games broadcast on national TV. It details traffic patterns in Germany during the opening match, noting significant drops during the game and brief recoveries during halftime. The post also reports on application-layer DDoS attacks targeting websites in Germany during the event, including a significant attack on a translation tool that peaked at 1.74 million requests per second.
  34. Exam-ining recent Internet shutdowns in Syria, Iraq, and AlgeriaThis post details the impact of recent internet shutdowns in Syria, Iraq, and Algeria, implemented to prevent exam cheating. It analyzes how Syria uses BGP to effectively shut down internet access by not announcing IP address space, leading to asymmetrical traffic flow. The post also examines the impact on DNS resolution and application traffic, and contrasts Syria's approach with Algeria's content blocking strategy. This contributes to the understanding of how national entities leverage network infrastructure for control and the resulting implications for global internet connectivity.
  35. Undersea cable failures cause Internet disruptions for multiple African countriesThis post details the impact of multiple undersea cable failures on internet connectivity in several African countries on March 14, 2024. It leverages Cloudflare Radar data to map the geographical spread and temporal progression of disruptions, identifying specific countries and network providers affected. The post highlights the reliance of the internet on submarine cables and references previous similar incidents. It also details the specific cables impacted (WACS, MainOne, SAT-3, ACE) and mentions the impact on Microsoft Azure and the MainOne website. The post concludes with a commitment to continued monitoring and provides links to Cloudflare Radar's Outage Center and social media channels for updates.
  36. Network performance update: Security Week 2024This post provides an update on Cloudflare's network performance benchmarking, detailing improvements in 95th percentile TCP connection times and highlighting the impact of HTTP/3 adoption. It introduces a refined methodology for network selection, ensuring a consistent set of top 1000 networks for measurement. The post quantifies Cloudflare's performance advantage over competitors like Amazon CloudFront, Google, Akamai, and Fastly, showing it as the fastest provider in 44% of measured networks for TCP connection time and 73% for mean TCP connection time. It explains how HTTP/3, built on QUIC, accelerates connection setup by enabling parallel TLS encryption and connection establishment, leading to significant latency reductions.
  37. Simplifying how enterprises connect to Cloudflare with Express Cloudflare Network InterconnectThis post announces 'Express CNI,' a significant update to Cloudflare Network Interconnect (CNI). It introduces a drastically simplified and accelerated provisioning process (under 3 minutes and 3 clicks) directly from the Cloudflare dashboard. Key improvements include the elimination of GRE tunnels for Magic Transit and Magic WAN, enabling direct 1500-byte packet transfer. The post also highlights the cost benefits of CNI, emphasizing no port fees or inter-region bandwidth charges, and details the user experience of ordering and configuring a CNI. Future plans include extending Express CNI features to interconnection partners and expanding data center support.
  38. connect() - why are you so slow?This post investigates the performance implications of reducing the number of source IPv4 addresses used by Cloudflare's cache service for egress connections. It identifies a performance bottleneck in the TCP connect() syscall, specifically within the `__inet_hash_connect()` function, which is exacerbated when using a single IPv4 address due to port exhaustion. The analysis reveals that the Linux kernel's port selection algorithm, which prioritizes even or odd ports in separate passes, leads to significant latency when even ports are exhausted. This research directly contributes to Cloudflare's strategy of deprecating IPv4 and moving towards IPv6 by understanding and mitigating the performance challenges associated with reduced IPv4 address allocation.
  39. Using DNS to estimate the worldwide state of IPv6 adoptionThis post contributes to the understanding of Cloudflare's global network expansion by providing a detailed analysis of IPv6 adoption. It leverages DNS query data from 1.1.1.1 to estimate both client-side and server-side IPv6 adoption, comparing these findings with HTTP traffic data from Cloudflare Radar. The post highlights the current state of IPv6 adoption, its historical growth patterns, and the challenges and importance of its continued deployment.
  40. Internet traffic patterns in Israel and Palestine following the October 2023 attacksThis post leverages Cloudflare's global network data to provide real-time insights into internet traffic patterns, network outages, and cyberattacks in Israel and Palestine following the October 2023 attacks. It details specific network disruptions in the Gaza Strip, traffic surges in Israel, and significant DDoS attacks targeting both regions, showcasing the network's ability to monitor and report on critical events impacting internet connectivity and security on a global scale.
  41. Amazon’s $2bn IPv4 tax — and how you can avoid paying itThis post addresses the emerging challenge of AWS imposing charges for IPv4 addresses, framing it as an interruption to the internet's cost reduction trend. It positions Cloudflare's global network and its robust IPv6 support as a solution, enabling customers to transition to IPv6 with AWS and avoid these new costs. The post provides a step-by-step guide for configuring Cloudflare and AWS for IPv6 compatibility, and offers a financial credit to users who adopt this solution. It also elaborates on the technical and economic reasons behind the scarcity of IPv4 addresses and the advantages of IPv6.
  42. Traffic anomalies and notifications with Cloudflare RadarThis post introduces the integration of an internal traffic anomaly detection and alerting tool into Cloudflare Radar, making country and network-level traffic anomalies visible on the Outage Center and available via API. It also launches new notification functionality allowing users to subscribe to alerts for traffic anomalies, confirmed Internet outages, route leaks, or route hijacks. This enhances the network's ability to provide real-time insights into global internet health and disruptions.
  43. How Cloudflare’s systems dynamically route traffic across the globeIntroduced Traffic Manager, an automated system for dynamically routing traffic across Cloudflare's global network to balance supply and demand. Developed Traffic Predictor, a system that simulates data center failures to predict traffic failover paths and inform Traffic Manager's policies. This enables granular route adjustments to prevent issues like 'thundering herd' scenarios and 'stranding capacity'.
  44. Cloudflare Radar's new BGP origin hijack detection systemThis post introduces Cloudflare Radar's new BGP origin hijack detection system, bringing internal monitoring capabilities to the public. It details the mechanics of BGP and origin hijacking, explains the limitations of current prevention mechanisms like RPKI, and outlines the design of Cloudflare's detection system, which uses a prefix trie to monitor BGP update messages for origin changes and identify potential hijacks. This enhances the network's security and stability by providing public visibility into routing anomalies.
  45. Routing information now on Cloudflare RadarThis post introduces the Cloudflare Radar Routing page, which provides detailed insights into BGP messages, IP prefixes, autonomous systems, and countries. It enables users to examine routing status, identify anomalies such as route leaks, hijacks, and RPKI invalid MOASes, and understand network connectivity. The post also highlights the availability of API access to this routing data, further enhancing the capabilities for monitoring and analyzing the internet's routing infrastructure.
  46. Network performance update: Speed Week 2023This post details Cloudflare's ongoing efforts to improve network performance, specifically highlighting the increase in networks where Cloudflare is the fastest provider (from 33% to 46%) and the improvement in networks where it's within 2ms or 5% of the fastest (from 8% to 10%). It outlines the tooling and processes used, including Real User Measurements (RUM) and a Performance Benchmarks Dashboard, to monitor metrics like TCP Connection Time and Time to First Byte (TTFB). The post also details the three main categories of actions taken to improve performance: establishing peering, expanding compute capacity, and applying traffic engineering.
  47. How Orpheus automatically routes around bad Internet weatherThis post details the Orpheus system, which identifies Internet connectivity outages beyond Cloudflare's network in real time and leverages Cloudflare's global network to find alternative paths around these outages. It explains how Orpheus reduces 522 errors by ensuring origin reachability even when direct BGP routes fail or are compromised, and how it reuses components of Argo Smart Routing's control plane to provide failover paths. The post highlights Orpheus's impact, saving billions of requests and preventing numerous incidents.
  48. Cloudflare's global network grows to 300 cities and ever closer to end users with connections to 12,000 networksThis post announces the expansion of Cloudflare's global network to over 300 cities and connections to over 12,000 Internet networks. It highlights the strategic importance of this expansion for reducing latency, improving interconnections with local ISPs and ASNs, and building network resilience. Specific examples include improved latency in Brazil and Pakistan due to local interconnections. The post also details the significance of connecting to a large fraction of global ASNs for performance and resilience.
  49. Exam-related Internet shutdowns in Iraq and Algeria put connectivity to the testThis post contributes to the global network expansion feature by detailing how Cloudflare Radar monitors and reports on government-imposed internet shutdowns in Iraq and Algeria during exam periods. It analyzes traffic data, BGP announcements, and DNS resolver usage (1.1.1.1) to understand the nature and impact of these disruptions, highlighting differences in implementation (routing vs. blocking) and regional variations. This showcases Cloudflare's role in providing visibility into and understanding of internet stability challenges globally.
  50. Improving customer experience in China using China ExpressThis post details the launch of China Express, a new offering that enhances Cloudflare's China Network by providing optimized, high-quality circuits for overseas connectivity in partnership with China Mobile International (CMI), CBC Tech, and Niaoyun. It addresses performance bottlenecks and packet loss for dynamic content requests originating from mainland China by routing them through dedicated circuits to Hong Kong and then to origin servers. The post presents a real-world example of an online shopping platform that experienced significant improvements in average download throughput and load times after implementing China Express, demonstrating a 54% performance improvement compared to using the public internet for origin pull requests during peak hours.
  51. Cloudflare’s view of Internet disruptions in PakistanThis post details the impact of government-ordered internet shutdowns and social media blocking in Pakistan on network traffic, as observed through Cloudflare's data. It analyzes traffic trends at national, administrative unit, and city levels, highlighting the shift from mobile to fixed broadband, the increase in 1.1.1.1 resolver usage, and the observed increases in traffic to blocked social media platforms. This contributes to the understanding of how Cloudflare's global network infrastructure is affected by and can report on large-scale internet disruptions.
  52. The European Network Usage Fees proposal is about much more than a fight between Big Tech and Big European telcosThis post argues against proposed network usage fees in Europe, explaining how they would create fast lanes for Big Tech and slow lanes for others, negatively impacting European consumers and small businesses. It details Cloudflare's extensive peering infrastructure and advocates for settlement-free peering as a mechanism to keep the internet open and performant, contrasting it with the proposed regulatory intervention that would artificially inflate interconnection prices.
  53. Effects of the conflict in Sudan on Internet patternsThis post leverages Cloudflare's global network visibility and analysis tools (Cloudflare Radar) to document the significant internet disruption in Sudan caused by armed conflict. It details traffic drops, ISP outages, BGP update spikes, and shifts in messaging app usage, providing real-time data and insights into the impact of geopolitical events on internet infrastructure and user behavior. It also highlights the disproportionate reliance on mobile traffic in Sudan and its implications during such crises.
  54. Making home Internet faster has little to do with “speed”This post analyzes the impact of bandwidth and latency on internet performance, using real-world data and simulations. It demonstrates that latency has a more significant impact on web page load times than bandwidth beyond a certain threshold. The post explains how latency affects connection establishment (TCP/TLS/QUIC handshakes) and sequential loading of web assets. Cloudflare's strategy of expanding its global network to reduce latency is presented as a solution to improve Internet Quality of Experience (QoE).
  55. Internet disruptions overview for Q1 2023This post contributes to the Global Network Expansion thread by providing a detailed overview of internet disruptions observed in Q1 2023. It showcases Cloudflare's ability to monitor and report on these events across numerous countries and network providers, highlighting the impact of government-directed shutdowns, cable cuts, and power outages on internet stability. This reinforces the importance of Cloudflare's extensive global network in understanding and mitigating such disruptions.
  56. Cloudflare’s view of the Virgin Media outage in the UKThis post details how Cloudflare Radar observed and analyzed a significant outage experienced by Virgin Media (AS5089) in the UK. It provides concrete data on traffic drops and correlates them with BGP announcement and withdrawal activity, demonstrating the practical application of network monitoring and analysis in understanding and reporting on internet disruptions. The post also highlights the cascading effect of such outages, including the unavailability of the ISP's own status page due to its DNS nameservers being hosted within the affected network.
  57. Cloudflare's network expansion in IndonesiaThis post details Cloudflare's network expansion into Indonesia, highlighting the geographical, infrastructural, and market challenges. It outlines the specific network deployments made (PoP at NTT, partnership with CitranetIX, direct interconnections, peering, transport to Singapore) and presents performance benchmarks (connect time, TTFB) comparing Cloudflare's performance against other providers in Indonesia from late 2021 to late 2022, showing significant improvements and leadership in key networks.
  58. One year of war in Ukraine: Internet trends, attacks, and resilienceThis post details how Cloudflare's global network infrastructure and services played a critical role in maintaining internet resilience and connectivity in Ukraine during the first year of the war. It highlights the network's ability to absorb significant traffic shifts, mitigate application-layer cyberattacks, reroute traffic away from compromised networks, and support the increased use of satellite internet services like Starlink. The analysis demonstrates the network's robustness in the face of physical infrastructure damage and cyber warfare, showcasing its capacity to adapt to extreme geopolitical events and ensure continued internet access.
  59. Cloud CNI privately connects your clouds to CloudflareThis post announces the expansion of Cloudflare Network Interconnect (CNI) to support direct private connections from major public cloud providers (IBM Cloud, Google Cloud, Azure, OCI, AWS) into Cloudflare's network. This allows customers to manage their cloud connectivity privately, integrating seamlessly with Cloudflare One products like Magic WAN and Access, simplifying network architecture and reducing vendor dependencies. The post details the ease of provisioning these connections directly from cloud provider portals.
  60. China Express: Cloudflare partners to boost performance in China for corporate networksIntroduced China Express, a suite of offerings to improve performance and connectivity for corporate networks in mainland China. This includes Premium DIA (Dedicated Internet Access) with partners CMI and CBC Tech for optimized cross-border connectivity, Private Link for secure private network tunnels, and Travel SIM in partnership with CMI for Zero Trust security on mobile devices. This extends Cloudflare One's Zero Trust capabilities to China, enabling consistent security policies and Magic WAN integration.
  61. Helping build a safer Internet by measuring BGP RPKI Route Origin ValidationThis post introduces a new method for measuring the adoption of RPKI Route Origin Validation (ROV) by Internet Service Providers. It details the challenges of measuring ROV deployment and presents Cloudflare's novel approach using browser-based probes to continuously monitor the reachability of RPKI-invalid origins. The post provides global data on ROV deployment, highlighting significant variations by country and emphasizing the need for broader adoption to secure internet routing.
  62. How Cloudflare helps next-generation marketsThis post details how Cloudflare's Edge Partner Program (EPP) helps improve performance in next-generation markets by embedding nodes within ISP networks, shortening the distance to users and reducing latency, packet loss, and retransmissions. It highlights the strategic expansion into 80% of new cities being in next-generation markets over the past four years to improve cached and dynamic performance. The post also mentions how Argo Smart Routing improves dynamic and uncached content performance by over 30%.
  63. Cloudflare expands Project Pangea to connect and protect (even) more community networksThis post announces the expansion of Project Pangea, a program that provides free internet connectivity, DDoS protection, and network firewalling to underserved community networks. The key expansion is the relaxation of technical requirements, specifically removing the need for participants to bring their own /24 IP space, making the program accessible to a wider range of community networks. It highlights the impact of Magic Transit and Magic Firewall in these networks and shares a success story from Ayva Networks.
  64. The Montgomery, Alabama Internet Exchange is making the Internet faster. We’re happy to be there.This post highlights Cloudflare's participation in the Montgomery, Alabama Internet Exchange (MGMix). It explains the role of IXs in improving internet performance by facilitating direct traffic exchange between networks, reducing reliance on costly transit providers. The post details the challenges and successes of establishing MGMix, including its impact on reducing latency and improving throughput in underserved regions. Cloudflare's decision to join MGMix is presented as a strategic move to enhance local internet performance and support digital equity.
  65. Cloudflare partners to simplify China connectivity for corporate networksThis post announces the expansion of Cloudflare One's SASE capabilities into mainland China through strategic partnerships. It details how these partnerships enable accelerated traffic from China to resources outside China by optimizing cross-border packet flows and enforcing security policies. It also describes how the WARP client will be adapted to enforce uniform security policies for remote users in China, and outlines future plans to extend full SASE filtering to local China data centers.
  66. Cloudflare servers don't own IPs anymore – so how do they connect to the Internet?This post details the evolution of Cloudflare's egress network design. It explains the challenges of using anycast for egress traffic and the shift from a unicast IP per server model to a more complex system. The post introduces the need for differentiating traffic types (trusted vs. untrusted, country-tagged) and the resulting explosion in required IP addresses. It then discusses two extreme solutions (each server owns all IPs vs. one server owns IPs) and proposes a third, hybrid way that leverages IP address management and routing strategies to handle diverse traffic types while maintaining the benefits of anycast for ingress traffic. This includes strategies for managing IP address cardinality and cost, particularly for IPv4.
  67. Why BGP communities are better than AS-path prependsThis post explains the fundamental concepts of BGP, including Autonomous Systems, IP prefixes, and routing announcements. It details how BGP messages carry metadata like AS-path and communities, and introduces the BGP best path selection algorithm. The post specifically contrasts AS-path prepending with BGP communities as traffic engineering tools, explaining the limitations of prepending and the advantages of communities for influencing inbound traffic, showcasing Cloudflare's advanced understanding and application of these protocols for network optimization and resilience.
  68. How we detect route leaks and our new Cloudflare Radar route leak serviceThis post introduces Cloudflare Radar's route leak data and API, making information about route leaks across the internet publicly available. It details the system built to detect route leaks using data from public sources and Cloudflare's global network, and explains the concepts of BGP and various types of route leaks (Type 1, 2, 3, and 4) as defined by RFC7908. This enhances the 'Global Network Expansion' thread by demonstrating how Cloudflare leverages its network visibility to actively monitor and mitigate internet routing issues, contributing to overall internet stability.
  69. Making peering easy with the new Cloudflare Peering PortalIntroduced a new Cloudflare Peering Portal that supports PeeringDB logins, eliminating the need for separate Cloudflare-specific accounts. This allows network operators to use their existing PeeringDB credentials to log in and manage peering sessions. Additionally, the portal now allows users to directly request peering sessions at Internet Exchanges, which are then processed by Cloudflare's network team. This significantly simplifies the process of establishing direct connections with Cloudflare's network.
  70. Internet disruptions overview for Q3 2022This post details observed internet disruptions during Q3 2022, categorizing them by cause (government-directed shutdowns) and geography. It provides specific examples from Iraq, Cuba, Afghanistan, Sierra Leone, Somaliland, India, and Iran, illustrating the impact of these shutdowns on internet resilience and highlighting Cloudflare's role in observing and reporting these events through its global network and tools like Cloudflare Radar.
  71. Protests spur Internet disruptions in IranThis post details how Cloudflare's network infrastructure and data analysis capabilities, particularly through Cloudflare Radar, were used to monitor and report on widespread internet disruptions and DNS blocking implemented by the Iranian government during protests. It provides concrete data on traffic drops across major Iranian ISPs and social media platforms, illustrating the impact of these government actions on internet accessibility and censorship.
  72. Regional Services comes to India, Japan and AustraliaThis post announces the general availability of 'Regional Services' in India, Japan, and Australia, expanding Cloudflare's ability to meet data localization requirements globally. It details how Regional Services work by leveraging Anycast and BGP to ensure traffic is routed to data centers within a customer-defined region for decryption and inspection, while raw traffic can be forwarded encrypted to the correct region if initially routed elsewhere. This directly addresses customer needs for regulatory compliance and contractual agreements regarding data handling.
  73. 1.1.1.1 + WARP: More features, still privateIntroduced a mechanism to provide geographically accurate browsing experiences for WARP users by replacing their original IP addresses with Cloudflare egress IPs that are associated with approximate locations in geolocation databases. This involves analyzing network traffic to estimate user distribution per city, allocating egress IPs using anycast, and submitting geolocation data to providers. Additionally, WARP+ subscribers now benefit from an expanded network with over 20 new cities.
  74. Cloudflare deployment in GuamThis post details the deployment of a new Cloudflare data center in Hagatna, Guam. It explains the geographical relevance of Guam for reducing latency to its residents and nearby archipelagos, highlights the submarine cables serving the region, and describes Cloudflare's peering with local ISPs via the Mariana Islands Internet Exchange (MARIIX). The post quantifies the performance impact with metrics showing significant reductions in Round Trip Time (RTT) for local users and notes an increase in HTTP requests in Guam post-deployment.
  75. Cloudflare’s view of the Rogers Communications outage in CanadaThis post details Cloudflare's analysis of the Rogers Communications outage in Canada, providing real-time data and insights into the event. It explains the role of BGP in internet routing and how a spike in BGP updates and prefix withdrawals led to the widespread disruption. The post showcases Cloudflare Radar's ability to monitor and report on such large-scale network events, emphasizing the interconnected nature of the internet and the impact of failures within major ISPs.
  76. Internet disruptions overview for Q2 2022This post contributes to the Global Network Expansion thread by providing a detailed overview of internet disruptions observed in Q2 2022. It showcases Cloudflare's ability to monitor and analyze network resilience through real-world events like optic outages (fiber cuts) and government-directed shutdowns. The post uses traffic graphs from Cloudflare Radar and internal tools to illustrate the impact of these disruptions on various regions and networks, including shifts in traffic composition (mobile/desktop, IPv4/IPv6). This data directly informs the understanding of network stability and the challenges faced in maintaining global internet performance.
  77. Network performance update: Cloudflare One Week June 2022This post details the continued expansion and performance improvements of Cloudflare's global network. It highlights an increase in the number of networks where Cloudflare is the fastest provider, from 1,280 to 1,290 out of the top 3,000 reported networks between May and June 2022. It also shows an increase in countries where Cloudflare is the fastest provider, with specific gains in Eastern Europe. The post emphasizes the importance of network performance for Zero Trust architectures and outlines ongoing efforts in expanding peering arrangements and adding new locations to maintain high performance.
  78. Cloudflare outage on June 21, 2022This post details a significant outage on June 21, 2022, caused by a misconfiguration during a BGP policy change in Cloudflare's newly architected Multi-Colo PoP (MCP) data centers. The change inadvertently withdrew critical site-local prefixes, leading to a loss of connectivity to 19 major data centers and impacting 50% of global traffic. The post provides a detailed timeline, technical description of the error (re-ordering of BGP policy terms causing an explicit reject), and outlines remediation steps focusing on process improvements, architectural redesigns, and enhanced automation for staggered rollouts and automated rollback.
  79. AAE-1 & SMW5 cable cuts impact millions of users across multiple countriesThis post details the impact of submarine cable cuts on the AAE-1 and SMW-5 cables, affecting millions of users across the Middle East, Africa, and Asia. It uses Cloudflare Radar data to illustrate traffic drops and recovery, and explains how Cloudflare's distributed network and proactive measures like Orpheus help maintain content availability and performance during such disruptions. It also highlights the importance of network redundancy and diverse physical paths for critical infrastructure.
  80. HTTP RFCs have evolved: A Cloudflare view of HTTP usage trendsThis post details the evolution of HTTP protocols, specifically highlighting the adoption and usage trends of HTTP/3 across Cloudflare's network. It analyzes traffic data from browsers and bots, showcasing the increasing adoption of HTTP/3 and its impact on network performance, reinforcing the network's role in supporting and driving the adoption of new internet standards.
  81. Tracking shifts in Internet connectivity in Kherson, UkraineThis post details how Cloudflare's global network and routing analysis capabilities were applied to track an internet outage and a shift in routing control in Kherson, Ukraine, during a conflict. It demonstrates the network's ability to observe and adapt to geopolitical disruptions, rerouting traffic to different data centers (e.g., Moscow) in response to upstream provider changes, and then reverting as connectivity was restored through local networks. This highlights the network's resilience and the importance of real-time monitoring in dynamic and challenging environments.
  82. The Cloudflare network now spans 275 citiesCloudflare announced the expansion of its global network to 275 cities, adding Ahmedabad, India; Chandigarh, India; Jeddah, Saudi Arabia; and Yogyakarta, Indonesia in the last month. The post details the performance improvements in India through interconnections with major ISPs, mobile network providers, and IXPs, stating that Cloudflare is connected to networks accounting for over 95% of the country's broadband subscribers. It highlights a 26% median performance improvement in Jeddah, reducing latency from 81ms to 60ms, which translates to approximately 55 hours saved per day for nearly 10 million requests.
  83. Internet disruptions overview for Q1 2022This post reviews selected Internet disruptions observed by Cloudflare during the first quarter of 2022, grouping them by associated cause. It details specific incidents in Tonga, Japan, The Gambia, Yemen, Tasmania, Iran, Cuba, Venezuela, Pakistan, Kazakhstan, Uzbekistan, Kyrgyzstan, and Taiwan, providing traffic graphs and attributing outages to volcanic eruptions, earthquakes, submarine cable damage, data center fires, and power outages. It also highlights the role of Cloudflare Radar and internal tools in observing these events.
  84. Cloudflare Radar’s new ASN pagesThis post introduces Cloudflare Radar's new Autonomous System (AS) pages, providing detailed traffic statistics, protocol usage, security information (DDoS attacks), geographical traffic distribution, and BGP announcement volumes for individual ASes. It highlights how this data can be used to understand unusual internet events like outages and attacks, and reinforces Cloudflare's commitment to increasing transparency into the internet's infrastructure.
  85. Network performance update: Security WeekThis post details Cloudflare's continuous efforts to improve network performance and expand its reach, reporting that it is now the fastest provider in 71% of the top 1,000 most reported networks globally. It highlights specific case studies in Canada and Australia where performance issues were diagnosed and resolved through network optimizations, including addressing congestion via improved peering and direct interconnection. The post also presents updated performance metrics showing improvements since the 'Full Stack Week' update, with Cloudflare becoming the fastest provider in 19 new networks and achieving #1 status in all of South America and several other countries.
  86. New cities on the Cloudflare global network: March 2022 editionThis post announces the addition of 18 new cities to Cloudflare's global network, bringing the total to over 270. It details the ISP Edge Partnership Program and provides a specific case study in Israel where new deployments reduced median response time by 66% (from 86ms to 29ms) by serving requests locally within the ISP's network instead of routing them to Europe. The post also lists the new cities and countries added.
  87. BGP security and confirmation biasesThis post details a specific BGP route leak incident caused by a configuration error during a network upgrade. It highlights the operational challenges of hot-cuts, the role of prefix limits as a safeguard, and the importance of BGP security protocols like RPKI and RFC8212. It also emphasizes the impact of confirmation bias on operational decision-making and the need for further network automation and training to prevent similar incidents.
  88. Internet is back in Tonga after 38 days of outageThis post details the restoration of internet connectivity to Tonga after a 38-day outage caused by an undersea cable break. It highlights Cloudflare Radar's role in monitoring the traffic increase post-repair and discusses the technical aspects of undersea cable systems, their reliability, and the specific incident involving the cable connecting Tonga to Fiji. It also touches upon the use of satellite services as a temporary alternative and the ongoing repairs for domestic cables.
  89. Internet shut down in Kazakhstan amid unrestThis post details Cloudflare's real-time monitoring and analysis of a nationwide internet shutdown in Kazakhstan, triggered by civil unrest. It uses Cloudflare Radar data to show the impact on internet traffic, mobile services, and BGP routing. The post highlights the correlation between internet service restoration and public announcements by the Kazakh President, suggesting a pattern of controlled unblocking. It also draws parallels to other internet shutdown events globally, underscoring Cloudflare's role in observing and reporting on network disruptions.
  90. How The Gambia lost access to the Internet for more than 8 hoursThis post analyzes the January 4, 2022, internet outage in The Gambia, detailing its cause as a failure in the ACE submarine cable and a subsequent issue with terrestrial backup links via Senegal's Sonatel. It uses BGP data to illustrate The Gambia's reliance on a single network path for international connectivity and highlights the critical nature of submarine cable systems and terrestrial interconnections for national internet access. The post also references The Gambia's regulatory authority's plans to explore securing a second fiber cable backup.
  91. How Cloudflare Is Solving Network Interconnection for CIOsThis post details Cloudflare's strategy for solving network interconnection challenges for CIOs and IT teams. It outlines three key areas: making interconnection valuable by providing access to security, reliability, and performance products; expanding interconnection availability through Cloudflare for Offices and partner networks; and reducing provisioning times by automating BGP configurations, streamlining cross-connect provisioning, and improving uptime. The post emphasizes the benefits of direct, private, and performant access to Cloudflare services, bypassing the public internet, and highlights the network effects derived from its large customer base.
  92. Sudan was cut off from the Internet for 25 daysThis post details a significant internet shutdown in Sudan lasting 25 days, highlighting the impact of political turmoil on internet access. It uses Cloudflare Radar data to show the restoration of traffic, the role of specific ISPs, and the prevalence of mobile traffic in the country. It also contextualizes this event within a history of internet disruptions in Sudan and emphasizes the commonality of such events globally.
  93. Network Performance Update: Full Stack WeekThis post details significant advancements in Cloudflare's network performance, reporting that Cloudflare has become the fastest provider in 79 new networks (a 14% increase) during Full Stack Week. It also highlights a 196% speed advantage for Cloudflare Workers over Fastly's Compute@Edge. The post provides specific case studies of network performance improvements in Peru and Sri Lanka, detailing how issues with traffic routing and latency were diagnosed and resolved, leading to Cloudflare achieving number one ranking in those regions. The post reinforces the ongoing commitment to continuous network optimization and benchmarking.
  94. Fixing Recent Validation Vulnerabilities in OctoRPKIThis post addresses recent vulnerabilities discovered in RPKI validation software, specifically within Cloudflare's OctoRPKI tool. It details the nature of these vulnerabilities (arbitrary file writes and uncontrolled resource consumption) and the mitigations implemented in version 1.4.0. The post reinforces Cloudflare's commitment to securing BGP routing through RPKI and highlights their proactive approach to addressing security research in this critical area.
  95. Sudan: seven days without Internet access (and counting)This post details a significant internet outage in Sudan, highlighting Cloudflare Radar's ability to monitor and report on country-wide internet access disruptions. It provides historical context for internet shutdowns in Sudan and emphasizes Cloudflare's role in observing global internet traffic and country-specific connectivity.
  96. Sudan woke up without InternetThis post leverages Cloudflare Radar to track and analyze the widespread internet outage in Sudan following a political coup. It provides data-driven evidence of the disruption to both mobile and desktop traffic across major ISPs, highlighting Cloudflare's ability to monitor and report on global internet connectivity events.
  97. Understanding how Facebook disappeared from the InternetThis post provides an external perspective on the Facebook outage, detailing how Facebook's internal BGP and DNS configuration changes led to their services becoming unreachable. It explains the role of BGP in internet routing, shows how Facebook stopped announcing its routes, and the subsequent impact on DNS resolution. The post also highlights Cloudflare's monitoring capabilities, its DNS resolver's resilience during the event, and the cascading effects on other internet services, underscoring the interconnectedness of the internet and the importance of robust network infrastructure.
  98. Two Weeks Later: Finding and Eliminating Long Tail LatenciesThis post details Cloudflare's efforts to improve network performance in 61 country/network pairs where latency exceeded 100ms behind the leading provider. It explains how Cloudflare identified these issues through benchmarking and then implemented specific technical solutions, including rectifying route advertisement in Brazil by rerouting traffic from a transit provider to a private network interconnect (PNI) via their backbone, establishing new peering sessions in Spain to keep traffic local, and optimizing upstream selection in Africa by adjusting BGP LOCAL_PREF values to prefer a transit provider over a private peer for specific network prefixes. The post quantifies the improvements, showing Cloudflare becoming the #1 network in 29 of these pairs and significantly improving performance in almost all of them.
  99. Project Myriagon: Cloudflare Passes 10,000 Connected NetworksThis post details the achievement of connecting to over 10,000 networks through Project Myriagon, a significant expansion of Cloudflare's global network. It explains the importance of Autonomous Systems (ASNs) and BGP in internet routing, highlighting how direct interconnections reduce latency and avoid congestion. The post quantifies the impact with a real-world example of a 90% reduction in median end-user latency. It also introduces Project CatchG, aiming to further increase network connectivity.
  100. Benchmarking Edge Network Performance: Akamai, Cloudflare, Amazon CloudFront, Fastly, and GoogleThis post introduces a detailed benchmarking experiment using Real User Measurement (RUM) data to compare Cloudflare's network performance against key competitors (Akamai, Amazon CloudFront, Fastly, Google). It outlines the methodology, including the use of the Resource Timing API and measurement of TCP connection time, TTFB, and TTLB across approximately 21,000 ASNs. The results demonstrate Cloudflare's leading performance in most global network segments and metrics, highlighting the company's commitment to continuous optimization and data-driven improvement of its global network.
  101. Magic makes your network fasterThis post details how Cloudflare's Magic suite (Magic Transit, Magic WAN, Magic Firewall) leverages the company's global network expansion and architecture to significantly improve network performance. It explains how Anycast, ubiquitous service deployment, and advanced routing techniques (beyond traditional BGP) reduce latency and eliminate the 'trombone effect' inherent in legacy network architectures. The post provides concrete examples and test results demonstrating performance gains, and introduces Argo Smart Routing for Packets as a further optimization.
  102. Cloudflare Backbone: A Fast Lane on the Busy Internet HighwayThis post details the construction and benefits of Cloudflare's dedicated backbone network, comprising long-distance and metro fiber optic cables connecting data centers globally. It explains how this backbone provides guaranteed network capacity and consistent latency, enabling faster and more reliable packet routing. The post includes a real-world test demonstrating performance improvements, showing an average response time reduction of 31% for requests from São Paulo and 44% from Tokyo when using the backbone compared to the public internet. It also highlights the integration of the backbone with dynamic routing technologies like Argo Smart Routing, Argo Tiered Caching, WARP+, and Orpheus to optimize route selection.
  103. Cloudflare Passes 250 Cities, Triples External Network Capacity, 8x-es BackboneThis post announces the expansion of Cloudflare's global network to over 250 cities, including 17 new cities outside mainland China and 37 cities in mainland China via a partnership with JD Cloud. It also details a 3.5x increase in external network capacity since early 2020, with 70% of this capacity being direct interconnects (PNIs/CNIs) or peering. Furthermore, the internal long-haul backbone network capacity has increased by over 800% since the start of 2020, connecting data centers with dedicated fiber and wavelength capacity. The post highlights the performance benefits of this expansion, such as reduced latency and improved connection times, and emphasizes the security advantages of keeping traffic on-network.
  104. Announcing Project Pangea: Helping Underserved Communities Expand Access to the Internet For FreeCloudflare is launching Project Pangea, offering its network services (Cloudflare Network Interconnect, Magic Transit, Magic Firewall) for free to nonprofit community networks serving underserved communities. This initiative aims to reduce the cost of Internet access for these communities by providing secure, performant, and reliable connectivity, along with DDoS mitigation and network-layer firewall services.
  105. Empowering customers with the Bandwidth AllianceThis post introduces and details the Bandwidth Alliance, a program launched in late 2018 with over fifteen cloud providers. The alliance aims to combat cloud vendor lock-in by reducing or eliminating egress fees for data transfer between Cloudflare and its partners. It explains how Cloudflare's extensive global network and peering capabilities enable these cost reductions through private network interconnects (PNI) and peered connections. The post highlights significant customer benefits in terms of cost savings and increased choice, providing specific examples and customer testimonials, and also mentions the integration with Microsoft Azure for discounted egress.
  106. Upgrading the Cloudflare China Network: better performance and security through product innovation and partnershipThis post details the significant upgrade of Cloudflare's China network through a partnership with JD Cloud. It highlights improvements in DNS serving by bringing nameservers within Mainland China, reducing latency and improving reliability. It also describes advancements in DDoS mitigation, pushing L7 attack detection and mitigation to the edge and integrating with local ISPs for L3/L4 attack mitigation. The post provides data demonstrating at least a 30% performance improvement in latency and TTFB for both static and dynamic content, as well as DNS response times, due to the enhanced network design and JD Cloud's infrastructure.
  107. Expanding Cloudflare to 25+ Cities in BrazilCloudflare has expanded its network to over 25 cities in Brazil through a partnership with a large ISP. This expansion aims to significantly improve internet experience for Brazilians by reducing latency and increasing throughput. The 25th-percentile latency for non-bot traffic has dropped from the mid-20ms range to sub-10ms. The expansion is part of Cloudflare's mission to improve internet access globally and is expected to increase Cloudflare's presence in Brazil by 12-16 times.
  108. Cloudflare customers can now use Microsoft Azure Data Transfer Routing Preference to enjoy lower data transfer costsThis post introduces a new integration with Microsoft Azure's Routing Preference service, enabling Cloudflare customers to benefit from lower data transfer costs when using Azure as an origin. It highlights the technical details of the interconnection, including private peering and internet exchanges, and outlines the simple steps for customers to enable this feature. The post also emphasizes the shared philosophy between Cloudflare and Microsoft in promoting customer choice and avoiding vendor lock-in.
  109. Cloudflare’s Network Doubles CPU Capacity and Expands Into Ten New Cities in Four New CountriesThis post announces the addition of ten new Cloudflare cities across four new countries (Georgia, Costa Rica, Tunisia, Myanmar) and the doubling of computational capacity since the start of pandemic-related lockdowns. It details the specific new cities and provides performance metrics (25th percentile latency) to demonstrate the impact of these network expansions on user experience, particularly in Kenya. The post also mentions ongoing experimentation with ARM, ASICs, and Nvidia technologies.
  110. Interconnect Anywhere — Reach Cloudflare’s network from 1,600+ locationsThis post announces the expansion of Cloudflare Network Interconnect (CNI) to over 1,600 locations through new and expanded partnerships with Layer 2 service fabric providers like PacketFabric, Megaport, Equinix Fabric, Console Connect, CoreSite Open Cloud Exchange, and Infiny by Epsilon. It details how customers can now connect to Cloudflare's network privately and securely from their existing fabrics, even when not in the same data center as Cloudflare. The post highlights the ease of provisioning these connections via partner portals and the benefits of reduced latency, increased redundancy, and improved security.
  111. QUIC Version 1 is live on CloudflareThis post announces the general availability of QUIC version 1 (RFC 9000) on Cloudflare, marking a significant milestone in the evolution of internet transport protocols. It details the technical aspects of enabling QUIC and HTTP/3, including the use of UDP, default security, faster handshakes, and the role of HTTP Alternative Services (Alt-Svc) for discovery. The post highlights Cloudflare's in-house open-source implementation, 'quiche', and its widespread adoption by other projects, underscoring Cloudflare's role in driving the standardization and deployment of this critical internet technology. It also provides insights into the current traffic share of QUIC/HTTP/3 and the tools used for testing and interoperability.
  112. Protecting Cloudflare Customers from BGP Insecurity with Route Leak DetectionThis post introduces 'Route Leak Detection,' a new network alerting feature for Cloudflare customers who have 'brought their own IP' (BYOIP) addresses. It explains the dangers of BGP route leaks and hijacks, how the detection feature works by monitoring global routing tables for unauthorized prefix advertisements, and how customers can configure it. The post also emphasizes the importance of RPKI for preventing such leaks and highlights Cloudflare's advocacy and ongoing efforts to improve internet routing security.
  113. Announcing Network On-ramp Partners for Cloudflare OneThis post announces the expansion of Cloudflare's global network reach through new 'Network On-ramp Partnerships'. It introduces partnerships with VMware, Aruba, and Infovista for WAN/SD-WAN connectivity, and with Digital Realty, CoreSite, EdgeConneX, 365 Data Centers, BBIX, Teraco, and Netrality Data Centers for private interconnection. These partnerships aim to simplify customer connections to Cloudflare's network from their existing infrastructure, enhancing security, performance, and observability for corporate networks and enabling a more seamless transition from traditional MPLS links.
  114. Uganda's January 13, 2021 Internet Shut DownThis post details Cloudflare's real-time monitoring and reporting capabilities during a government-ordered internet shutdown in Uganda. It showcases how Cloudflare Radar and network traffic analysis, including BGP routing data, were used to observe and document the complete cessation and subsequent re-establishment of internet connectivity in the region, highlighting the global reach and observational power of Cloudflare's network.
  115. Internet traffic disruption caused by the Christmas Day bombing in NashvilleThis post details the impact of the Nashville bombing on internet traffic, specifically observing the outage and subsequent recovery of AT&T services in the region through Cloudflare's traffic flow data. It demonstrates the network's ability to monitor and report on large-scale, localized disruptions caused by physical infrastructure damage, highlighting the resilience of services operating on battery power and the eventual return to normal traffic flows.
  116. The Internet is Getting Safer: Fall 2020 RPKI UpdateThis post provides a significant update on Cloudflare's efforts to secure internet routing via RPKI. It details the progress in ROA creation, the success of the 'isbgpsafeyet.com' website in driving adoption of BGP Origin Validation, and highlights the release of new versions of Cloudflare's open-source RPKI software (OctoRPKI and GoRTR). It also showcases the growth in usage of Cloudflare's public RPKI services and dashboard, and provides projections for future RPKI adoption.
  117. August 30th 2020: Analysis of CenturyLink/Level(3) outageThis post details how Cloudflare's diverse network of providers and automated systems successfully rerouted traffic around a significant CenturyLink/Level(3) outage, minimizing customer impact. It highlights the challenges posed by single-homed connectivity and BGP route withdrawal issues, and speculates on the root cause being a Flowspec rule impacting BGP announcements, drawing parallels to Cloudflare's own past Flowspec-induced outage. The post underscores the importance of robust network resilience and the complexities of internet routing protocols.
  118. Cloudflare Network Interconnection partnerships launchIntroduced the Cloudflare Network Interconnection Partner Program, enabling customers to connect privately and securely via software-defined layer 2 networking technology in 23 metro areas across five continents. This program offers secure, efficient, performant, and cost-effective connectivity options through partners like Console Connect, Megaport, PacketFabric, Equinix ECX Fabric, and Zayo Tranzact, reducing overhead and providing data center independence.
  119. Introducing Cloudflare Network InterconnectThis post introduces Cloudflare Network Interconnect (CNI), a new capability that allows customers to directly connect their on-premise networks to Cloudflare's global network. CNI offers private network interconnects (PNIs) and Internet Exchange (IX) options, providing enhanced security, reliability, and performance benefits compared to using the public internet. It integrates with existing Cloudflare products like Cloudflare Access, CDN, Magic Transit, and Cloudflare Workers, enabling dedicated on-ramps to Cloudflare's network for improved performance and reduced threat exposure, particularly for Magic Transit by removing GRE tunnels over the public internet. The post also provides a hypothetical example of Penguin Corp using CNI with Cloudflare for Teams and Magic Transit to secure its branch and core networks.
  120. Bringing Your Own IPs to Cloudflare (BYOIP)This post announces the general availability of Bring Your Own IP (BYOIP) for Cloudflare's Layer 7 products, Spectrum, and Magic Transit. It details how customers can use their own IP prefixes with Cloudflare's edge network, enabling them to benefit from Cloudflare's performance and security features on their existing IP space. The post highlights the administrative benefits for customers, the technical requirements for BYOIP, and the configuration options available through the UI and API, including support for regional deployments and RPKI.
  121. Cloudflare outage on July 17, 2020This post details a specific outage event caused by a configuration error in Cloudflare's backbone network. It explains the root cause: an incorrect BGP route leak due to a misconfigured policy-statement that removed a prefix-list condition, leading to all BGP routes being sent to all other backbone routers with a higher local-preference. This caused traffic meant for local compute nodes to be rerouted to Atlanta. The post also outlines the immediate actions taken, including disabling the affected router and restoring service, as well as future preventative measures: implementing a maximum-prefix limit on backbone BGP sessions and changing the BGP local-preference for local server routes.
  122. Cloudflare Network expands to more than 100 CountriesThis post details the expansion of Cloudflare's global network to 206 cities across more than 100 countries, including the addition of six new cities (Vientiane, Tegucigalpa, Johor Bahru, Monrovia, Bandar Seri Begawan, Paramaribo) and the completion of 40+ datacenter expansion projects. It also highlights the addition of over 1Tbps in dedicated backbone capacity. The post emphasizes the challenges and adaptations made in logistics and operations due to the COVID-19 pandemic, and the strategic importance of these new city deployments for improving security, performance, and reliability.
  123. Introducing Regional ServicesIntroduced 'Regional Services' allowing Enterprise customers to specify geographic regions for Layer 7 traffic processing (WAF, Workers, Bot Management). DDoS protection and network acceleration remain global. TLS keys are stored and used only within the chosen region. Initially supports US and EU regions.
  124. Creating a True One-Stop Solution for Companies to Go Global: Announcing a Partnership Between Cloudflare and JD Cloud & AIThis post announces a strategic partnership with JD Cloud & AI, significantly expanding Cloudflare's presence in mainland China by adding 150 data centers. This partnership aims to create a one-stop solution for companies operating globally, particularly those doing business in or out of China, by combining Cloudflare's international network with JD Cloud & AI's domestic infrastructure. It addresses the technical and logistical challenges of China's balkanized network architecture and regulatory environment, enabling seamless deployment of Cloudflare's security and performance services within China.
  125. Internet performance during the COVID-19 emergencyThis post provides an update on internet performance during the COVID-19 emergency, confirming Cloudflare's network capacity to handle increased traffic. It presents data on global internet usage growth by country and analyzes TCP Round Trip Time (RTT) to assess network congestion. The data shows that while some regions experienced increased RTT, indicating strain, overall internet performance remained stable, highlighting the resilience of the internet infrastructure.
  126. Is BGP Safe Yet? No. But we are tracking it carefullyThis post addresses the critical issue of BGP leaks and hijacks, highlighting the ongoing need to make BGP safer. It details Cloudflare's deployment of RPKI for origin validation, the release of isBGPSafeYet.com to track RPKI adoption by major networks, and the use of test prefixes to demonstrate RPKI effectiveness. The post emphasizes the importance of community-wide RPKI adoption to mitigate routing security threats.
  127. Remote work, regional lockdowns and migration of Internet usageThis post details how Cloudflare leverages its global network of edge routers and flow record sampling to analyze aggregate traffic information and infer geospatial distribution of internet traffic volume. It uses geolocation databases like Maxmind to determine the rough location of traffic origins and visualizes this data using heatmaps to demonstrate the migration of internet usage from office-focused areas to residential areas during the COVID-19 pandemic.
  128. On the shoulders of giants: recent changes in Internet trafficThis post analyzes the impact of global events, specifically the COVID-19 pandemic, on internet traffic patterns. It details observed surges in traffic towards video conferencing, streaming, and news services, alongside a decrease in traffic from business and university networks. The post presents data visualizations showing traffic shifts in various countries (US, Netherlands, France, Italy, Hong Kong) and at major Internet Exchange Points (AMS-IX, LINX, DE-CIX, MXP-IX, HKIX). It quantifies the observed traffic increases, ranging from 10% to 40% depending on the region and government actions, and emphasizes the inherent robustness of the internet infrastructure in handling such dynamic load changes.
  129. RPKI and the RTR protocolThis post introduces and explains the Resource Public Key Infrastructure (RPKI) to Router Protocol (RTR), detailing how it works to secure the BGP routing system. It highlights Cloudflare's operational use of RPKI for signing routes and filtering inbound traffic, as well as its open-source contributions like the Cloudflare RPKI Toolkit and GoRTR. The post emphasizes the importance of RTR for communicating validated route information from RPKI validators to routers, discusses the protocol's specifications and security considerations, and introduces a new open-source tool, `rpki-rtr-client`, for testing the RTR protocol.
  130. An EPYC trip to Rome: AMD is Cloudflare's 10th-generation Edge server CPUThis post details Cloudflare's adoption of AMD EPYC 7642 CPUs for its 10th-generation edge servers, marking a significant hardware upgrade. It highlights the move to higher core density and improved power efficiency (Requests per Watt) by switching from dual-socket Intel Xeon Platinum 6162s to a single-socket AMD EPYC. The post presents extensive benchmarking results across cryptography, compression, and Go/LuaJIT workloads, demonstrating the performance benefits of the new AMD hardware for Cloudflare's compute-intensive software stack, directly contributing to the network's performance and efficiency.
  131. Cloudflare’s Gen X: Servers for an Accelerated FutureThis post introduces Cloudflare's tenth generation of servers, 'Gen X', highlighting a significant shift away from Intel to AMD EPYC processors for improved performance and cost efficiency. It details the server's compute, memory, storage, and networking components, emphasizing the focus on CPU efficiency for Cloudflare's compute-heavy workloads. The post also discusses the architectural decision to use standard commodity hardware and the flexibility it provides for network expansion and service deployment, reinforcing the core principles of the global network expansion strategy.
  132. A cost-effective and extensible testbed for transport protocol developmentThis post details the development of a cost-effective and extensible testbed for transport protocol development, specifically for QUIC and HTTP/3. It outlines the goals of such a lab, including aiding protocol development, debugging, performance comparison, and bottleneck identification. The post describes the lab configuration with Raspberry Pis, Intel x86 boxes, and traffic shapers, as well as test automation and visualization techniques. It also highlights the importance of testing on mobile devices and the integration of smartphones into the testbed to understand real-world performance issues across different operating systems and network conditions. This directly contributes to Cloudflare's ability to innovate and improve network protocols that underpin its global network.
  133. Cloudflare Expanded to 200 Cities in 2019This post announces the expansion of Cloudflare's global network to 200 cities, detailing the addition of seven new data centers in Chattogram & Dhaka (Bangladesh), Honolulu (Hawaii, USA), Adelaide (Australia), Thimphu (Bhutan), and St George's (Grenada). It highlights the strategic importance of these locations for improving security, performance, and reliability for Internet properties, particularly in underserved regions and remote areas. The post also notes the evolution from co-locating in key hubs to fanning out globally and partnering with local ISPs.
  134. Cloudflare Global Network Expands to 193 CitiesThis post details Cloudflare's expansion of its global network to 193 cities in 90+ countries, adding 20 new locations in 2019. It describes the evolution from targeting major Internet exchanges to partnering with local ISPs to bring infrastructure closer to users. The post also highlights the increase in computational capacity at existing edge locations and the growth in directly connected networks to over 8,000. It touches on the challenges of capacity planning for a global anycast network and the use of simulations to optimize server deployment, as well as the flexibility required in the supply chain for hardware deployment.
  135. Extending Cloudflare to On-Prem NetworksThis post introduces Cloudflare Magic Transit, a new service that extends Cloudflare's global edge network and its DDoS mitigation capabilities to on-premise networks. It leverages BGP and Anycast to announce customer IP address space globally, ingest traffic, and provide secure, performant IP connectivity. This significantly expands the application of Cloudflare's network infrastructure beyond traditional web applications to encompass entire enterprise networks.
  136. The deep-dive into how Verizon and a BGP Optimizer Knocked Large Parts of the Internet Offline MondayThis post provides a deep-dive analysis of a major BGP route leak event that occurred on June 24, 2019, caused by Verizon and a BGP optimizer. It details the technical investigation using RIPE NCC archived data to identify affected routes, the timing of the leak (approximately two hours), and the implications of invalid route propagation. The post highlights the importance of BGP communities like NO_EXPORT and the role of RPKI ROAs in preventing such incidents, directly contributing to the understanding and mitigation of global network stability issues.
  137. How Verizon and a BGP Optimizer Knocked Large Parts of the Internet Offline TodayThis post details a significant BGP route leak incident that caused widespread internet outages, impacting Cloudflare's services. It explains the technical cause involving a BGP optimizer and Verizon's insufficient filtering, highlighting the vulnerabilities in current BGP practices. The post strongly advocates for the adoption of routing security measures like RPKI, prefix limits, and IRR filtering, emphasizing Cloudflare's own implementation and urging other network operators, particularly Verizon, to adopt them to prevent future incidents.
  138. MultiCloud... flareThis post significantly expands the 'Global Network Expansion' thread by detailing how Cloudflare's global network infrastructure is leveraged for advanced multi-cloud capabilities. It introduces load balancing across different cloud providers and on-premise hardware, the Argo Tunnel technology for simplified and secure multi-cloud networking, and the application of access control policies within this distributed network. It also reiterates the reliability benefits derived from the global, anycast network architecture in a multi-cloud context.
  139. Argo and the Cloudflare Global Private BackboneIntroduced the Cloudflare Global Private Backbone, a network of private fiber connections linking Cloudflare data centers. This backbone provides enhanced control over end-to-end performance and capacity management, resulting in significantly reduced jitter (e.g., 536μs vs. 4ms between Chicago and Newark) and improved reliability compared to transit providers. The backbone is automatically integrated with Argo Smart Routing, allowing it to select optimal paths across both the public internet and the private backbone, further improving TTFB latency by an additional 16%.
  140. Cloudflare’s RPKI ToolkitThis post details Cloudflare's implementation of Resource Public Key Infrastructure (RPKI) to enhance the security of BGP routing. It introduces two open-source tools, GoRTR and OctoRPKI, developed to simplify RPKI deployment for network operators. The post explains the benefits of RPKI for users and website visitors, such as reduced risk of cache poisoning and route hijacks, and outlines the technical details and usage of these tools.
  141. Cloudflare Support for Azure CustomersThis post details Cloudflare's efforts to enhance its integration with Microsoft Azure. It announces the availability of an Azure application for Cloudflare Argo Tunnel in the Azure Marketplace, enabling secure, encrypted connections between origins and the Cloudflare edge. It also highlights Cloudflare's support for Azure's serverless static web hosting by providing guides for SSL issuance and renewal, and promotes the use of Cloudflare's 1.1.1.1 DNS resolver with Azure, including daily performance monitoring on the platform.
  142. Ten new data centers: Cloudflare expands global network to 165 citiesCloudflare has added ten new data centers in the United States (Charlotte, Columbus), Bahrain (Manama), Russia (St. Petersburg), Vietnam (Hanoi, Ho Chi Minh City), Pakistan (Karachi, Lahore, Islamabad), and Réunion (Sainte-Marie). This expansion brings the total number of cities to 165, with 46 new cities added in the current year. The post details the specific locations and their strategic importance for improving performance and security.
  143. How a Nigerian ISP Accidentally Knocked Google OfflineThis post details a specific incident of a route leak caused by a Nigerian ISP's network misconfiguration, which led to a significant outage for Google and other services. It explains the technical details of route leaks, Autonomous System (AS) numbers, and AS Paths, illustrating how trust without verification across multiple networks (MainOne, China Telecom CN2, TransTelecom) amplified the impact. The post also highlights Cloudflare's auto-remediation capabilities in detecting and mitigating such incidents and advocates for improved network filtering and best practices like RPKI and BCP-38 to enhance internet stability.
  144. Lagos, Nigeria - Cloudflare’s 155th cityCloudflare has established a new data center in Lagos, Nigeria, colocated with an undersea cable operator and connected to the IXPN and WAF-IX exchanges. This expansion allows for local content delivery, improved DDoS mitigation, and faster, more reliable connections for Nigerian users, shifting traffic from European data centers.
  145. Cloudflare’s network boosts security and performance for IBM Cloud customersThis post details the integration of Cloudflare's network services (DDoS mitigation, WAF, smart routing, load balancing) into IBM Cloud's new Cloud Internet Services (CIS) Enterprise offering. It highlights how Cloudflare's global network of over 150 PoPs, with a capacity of 20+ Tbps, is used to protect and accelerate IBM Cloud customers' applications. The integration allows IBM Cloud customers to manage these services through the IBM Cloud UI, simplifying operations. It also mentions the Bandwidth Alliance waiver for data transfer fees between IBM and Cloudflare servers for CIS Enterprise plan customers.
  146. Cloudflare Peering Portal - BetaIntroduced the Cloudflare Peering Portal (Beta) to provide ISPs and hosting providers with detailed statistics on traffic exchange with Cloudflare. This includes time series traffic statistics, peering session statistics, prefix data, and POP relative traffic weighting. The portal aims to help partners estimate potential transit cost savings, performance improvements, and bandwidth management benefits of peering or hosting Cloudflare nodes.
  147. A Tour Inside Cloudflare's G9 ServersThis post details the evolution of Cloudflare's server hardware with the introduction of the G9 generation. It highlights significant upgrades in CPU cores (from 12 to 192 per server), network interface speeds (from 10G to 25G), and storage (datacenter-grade SSDs). The focus on maximizing compute density, improving performance per watt, and enhancing I/O latency directly contributes to the network's ability to handle increasing traffic volumes and maintain high performance across its global infrastructure.
  148. Ulaanbaatar, MongoliaCloudflare has launched a new data center in Ulaanbaatar, Mongolia, marking city number 154 on its network. This expansion significantly reduces latency for customers and end-users within Mongolia by bringing Cloudflare's network closer to the country. The process involved negotiating data center space, securing bandwidth partners, ensuring PCI/DSS compliance, managing complex international logistics for hardware shipment, and deploying the infrastructure with a focus on network automation and security. The addition of Ulaanbaatar completes Cloudflare's network coverage for all letters of the alphabet.
  149. Bandwidth Alliance: powered by smart routing on Cloudflare’s networkIntroduced the Bandwidth Alliance, a collaboration between Cloudflare and other cloud providers to reduce data transfer fees for mutual customers. This alliance is technically underpinned by Cloudflare's Argo traffic routing engine, which is adapted to optimize for cost by leveraging direct peering and low-cost connections with alliance partners. A limited, routing-only version of Argo is provided to alliance partners to ensure traffic is routed to the nearest Cloudflare location, facilitating the use of Private Network Interfaces (PNI) and reducing bandwidth costs.
  150. Introducing the Bandwidth Alliance: sharing the benefits of interconnected networksIntroduced the Bandwidth Alliance, a partnership with cloud and networking companies to reduce or eliminate data transfer fees for mutual customers. This initiative leverages existing peered connections and introduces smart routing to ensure traffic between Cloudflare and participating providers qualifies for reduced or zero bandwidth rates. Founding members include Google Cloud, Microsoft Azure, IBM Cloud, Automattic, Backblaze, DigitalOcean, DreamHost, Linode, Packet, Scaleway, and Vapor. The estimated annual savings for Cloudflare customers is $50 million.
  151. RPKI and BGP: our path to securing Internet RoutingThis post details Cloudflare's proactive deployment of RPKI for securing BGP routing decisions, signing its own routes, and offering an RTR server to encourage wider adoption of route origin validation. It explains the technical underpinnings of BGP and RPKI, Cloudflare's implementation strategy, and the development of GoRTR to securely distribute RPKI caches. The post also highlights the importance of certificate transparency and ongoing efforts to combat route leaks and promote RPKI adoption.
  152. RPKI - The required cryptographic upgrade to BGP routingThis post introduces Resource Public Key Infrastructure (RPKI) as a cryptographic method to secure BGP routing, addressing the pervasive issues of route leaks and hijacks. Cloudflare announces its commitment to deploying RPKI and outlines the basic concepts of BGP, ASNs, and CIDRs, contrasting RPKI with the less secure Internet Routing Registry (IRR) system. It emphasizes RPKI's role in securing route origins and Cloudflare's leadership in promoting its adoption for global BGP security.
  153. Fixing an old hack - why we are bumping the IPv6 MTUThis post details the decision to increase the IPv6 MTU on Cloudflare's edge servers from a hardcoded 1280 bytes to 1400 bytes. This change was driven by the need to support UDP traffic forwarding for Cloudflare Spectrum, which requires a larger MTU to accommodate encapsulated VPN packets. The post explains the original problem with Path MTU discovery and ICMP packet drops in the ECMP routing setup, the temporary fix of reducing the MTU, and the analysis of eyeball MTU distributions to select a new, more appropriate MTU value that minimizes reliance on Path MTU Detection.
  154. African traffic growth and predictions for the futureThis post details Cloudflare's expansion of its datacenter presence in Africa, highlighting the growth of Internet usage on the continent and the impact of localized datacenters on latency and throughput. It analyzes traffic growth patterns in specific African countries after datacenter turn-ups, explains the relationship between latency and TCP throughput, and discusses inter-African routing challenges, often involving backhauling traffic through Europe. The post also visualizes latencies from Europe to African countries and analyzes common routing paths to Cloudflare datacenters, noting language and content provider biases. Finally, it briefly touches on the location of popular content providers within Africa.
  155. IPv6 in ChinaThis post details the rapid growth of IPv6 users and traffic originating from Mainland China, specifically highlighting China Mobile's aggressive IPv6 enablement and announcement of IPv6 address blocks to the internet. It discusses the necessity of IPv6 due to IPv4 exhaustion and the challenges of Carrier Grade NAT (CGN) and NAT64. The post notes that on June 7th, China Mobile started announcing IPv6 address blocks, and Cloudflare observed a significant increase in IPv6 traffic from China Mobile users.
  156. BGP leaks and cryptocurrenciesThis post details a significant BGP leak incident that affected Amazon's Route53 DNS servers and, consequently, Cloudflare's 1.1.1.1 DNS resolver in specific regions. It explains the mechanics of BGP leaks, how this particular incident was exploited to facilitate cryptocurrency theft via a phishing website, and the various actors and vulnerabilities involved. The post also highlights Cloudflare's role in detecting the leak and discusses potential solutions and existing protections like DNSSEC, HSTS, and DANE to mitigate such attacks.
  157. Fixing reachability to 1.1.1.1, GLOBALLY!This post details the challenges and efforts involved in ensuring global reachability to Cloudflare's new DNS resolver, 1.1.1.1. It highlights how widespread misconfigurations and intentional misuse of the 1.1.1.1 IP address by ISPs and CPE manufacturers were identified using RIPE Atlas probes. The post outlines the process of contacting these operators to resolve issues, leading to significant improvements in 1.1.1.1 availability worldwide. It also touches upon the unsolicited background traffic observed on these IPs, indicating potential botnet activity or misconfigurations.
  158. Welcome, Luxembourg City and Chișinău! Cloudflare Global Network Spans 151 CitiesThis post announces the addition of two new data centers in Luxembourg City and Chișinău, bringing the total to 151 cities globally. It details the process of turning up a new data center, including traffic evaluation, facility assessment, hardware logistics, physical installation, and remote provisioning. The post also highlights the rapid pace of network expansion, with 31 cities added in March 2018, and reiterates the aspiration to have 95% of the world's Internet users within a country with a Cloudflare data center by the end of 2018.
  159. Three more US data centers! Jacksonville, Memphis and TallahasseeAdded three new data centers in the Southern United States: Jacksonville (Florida), Memphis (Tennessee), and Tallahassee (Florida), bringing the total to 146th, 147th, and 148th cities globally, and 40th, 41st, and 42nd in North America. This expansion aims to reduce latency and increase capacity to withstand attacks.
  160. Riga, Tallinn and Vilnius: Launching three new European Cloudflare data centersCloudflare announces the turn up of three new data centers in Riga (Latvia), Tallinn (Estonia), and Vilnius (Lithuania), bringing the total to 142 cities globally and 39 in Europe. These new locations will improve security and performance for Internet properties in the Baltic states by interconnecting with local networks over multiple Internet exchanges.
  161. Cloudflare Global Network Spans 137 Cities: Launching Durban and Port Louis Data CentersThis post announces the launch of two new data centers in Durban, South Africa, and Port Louis, Mauritius, expanding the Cloudflare global network to 137 cities. It details the strategic importance of these locations for improving latency, providing redundancy, and withstanding DDoS attacks. The post also highlights ongoing expansion efforts in Africa and partnerships with ISPs.
  162. Mexico City, Mexico: Cloudflare Data Center #134This post announces the launch of Cloudflare's 134th data center in Mexico City, Mexico. This expansion marks Cloudflare's entry into Mexico and will serve traffic to over 7 million websites, improving performance for users in the Mexico City Metropolitan Area. This is the 10th data center in the Latin America and Caribbean region. The deployment was made possible through a partnership with Kio Networks, and Cloudflare will also be interconnecting with IXP Mexico. Future expansions are planned for Guadalajara, Monterrey, and Queretaro.
  163. Five new Cloudflare data centers across the United StatesCloudflare has added five new data centers in the United States: Houston (Texas), Indianapolis (Indiana), Montgomery (Alabama), Pittsburgh (Pennsylvania), and Sacramento (California), bringing the total to 133 data centers. These deployments improve performance, security, and reliability for customers by expanding the edge network. The company connects with local ISPs and participates in Internet Exchange Points in these new locations.
  164. Macau: Cloudflare Data Center 127Announced the launch of Cloudflare's 127th data center in Macau, China, which is their 44th data center in Asia. This expansion aims to make over 7 million Internet-facing applications safer and faster.
  165. Reykjavík, Cloudflare’s northernmost locationThis post announces the launch of Cloudflare's 125th global location in Reykjavík, Iceland. It details the strategic importance of this location for reducing latency for Icelandic users by connecting directly to the RIX Internet Exchange and leveraging submarine cables for global connectivity. The post also highlights Iceland's unique environmental conditions (green electricity, natural cooling) suitable for power-intensive compute operations.
  166. Istanbul (not Constantinople): Cloudflare’s 124th Data CenterThis post announces the launch of Cloudflare's 124th data center in Istanbul, Turkey. It highlights the strategic importance of this location, spanning two continents, and its significance for serving the Turkish market. The post also mentions the ongoing rapid expansion of Cloudflare's data center footprint globally.
  167. Kathmandu, Nepal is data center 123This post announces the launch of Cloudflare's 123rd data center in Kathmandu, Nepal. It highlights key facts about Nepal and Kathmandu, discusses the mountainous geography, and mentions the unique shape and symbolism of the Nepalese flag. It also notes the timing of the launch with the APRICOT 2018 conference and the significance of Nepal within the SAARC organization. The post concludes with a forward-looking statement about future data center deployments and a call for engineers to join the team.
  168. Marhaba Beirut! Cloudflare’s 121st location - مرحبا بيروت! موقع “كلاودفلار” ال ١٢١This post announces the deployment of Cloudflare's 121st global location in Beirut, Lebanon. It details the technical benefit of this deployment: reducing round-trip latency by approximately 50 milliseconds for internet traffic to and from Lebanon. This is achieved by connecting directly to Ogero, the sole network provider in Lebanon, rather than routing traffic through existing deployments in Marseille and Paris. The post also mentions the low internet speeds in Lebanon and the government's plan to improve connectivity with fiber optics.
  169. McAllen, Texas: Cloudflare opens 119th Data Center just north of the Mexico borderThis post details the launch of Cloudflare's 119th data center in McAllen, Texas, strategically located near the US-Mexico border. The primary technical driver for this deployment is to reduce latency for traffic destined for Mexico by approximately 12 milliseconds, by providing a local interconnect point for Mexican ISPs and telcos. The post also highlights the establishment of a new Internet Exchange (IX) in McAllen, MEX-IX, and Cloudflare's intention to enable peering on this IX to further improve interconnection with Mexican carriers.
  170. Living In A Multi-Cloud WorldThis post introduces the concept of multi-cloud strategies as a growing concern for companies, driven by the desire to avoid vendor lock-in. It highlights the benefits of multi-cloud, including cost efficiencies, negotiation strength, business continuity, and experimentation. Cloudflare's role is positioned as an independent enabler of multi-cloud strategies through its Load Balancing and Global Anycast Network. The post also provides a concrete example with Billforward, demonstrating how Cloudflare's Load Balancing facilitates failover between Google Compute Engine and Amazon Web Services, showcasing the ability to manage traffic across disparate cloud providers with minimal operational changes and no downtime.
  171. Dobar dan, Hrvatska! Announcing Cloudflare's Zagreb Data CenterAnnounced the launch of a new data center in Zagreb, Croatia, expanding the global network to 118 cities across 58 countries. This deployment aims to reduce latency for users in Croatia and the surrounding region by bringing Cloudflare's network closer to them, improving the security and performance of internet properties.
  172. Hurricane IrmaThis post details Cloudflare's preparedness and resilience measures for Hurricane Irma, highlighting the use of Tier IV data centers with robust power and cooling redundancy, on-site emergency supplies, and multiple connectivity options. It also explains how Anycast routing would reroute traffic to other US data centers in the event of a Miami datacenter outage, demonstrating the network's ability to withstand significant regional disruptions.
  173. Portland (Oregon): Cloudflare's 117th Data Center!This post announces the launch of Cloudflare's 117th global data center in Portland, Oregon. This deployment adds redundancy to existing data centers in Seattle and San Jose, increases capacity for services and attack mitigation, and expands the edge network closer to millions of Oregon Internet users. The Portland data center interconnects with ISPs like Comcast and the Northwest Access Exchange (NWAX). The company is actively expanding its global network with dozens of additional data centers planned.
  174. Broken packets: IP fragmentation is flawedThis post delves into the technical challenges of IP fragmentation and the importance of Path MTU Discovery for reliable internet communication. It highlights how Cloudflare's network infrastructure must contend with issues arising from intermediate network devices (routers, NATs, middle boxes) that can interfere with fragmentation and ICMP 'Packet too big' messages, impacting protocols like TCP. The post details various scenarios where these issues manifest and the complexities introduced by modern network architectures like ECMP and Anycast, underscoring the need for robust network engineering to ensure consistent performance and connectivity.
  175. Power outage hits the island of Taiwan. Here’s what we learned.This post highlights the resilience of Cloudflare's network during a major power outage in Taiwan. It demonstrates that even when local infrastructure fails, Cloudflare's data centers, with backup power, can continue to operate. The post also provides insights into user behavior during such events, noting an initial spike in requests as people sought information and a shift from desktop to mobile devices, leading to a decrease in bandwidth usage. This showcases the robustness of the global network in the face of localized disruptions.
  176. Aquele Abraço Rio de Janeiro: Cloudflare's 116th Data Center!This post announces the launch of Cloudflare's 116th data center in Rio de Janeiro, Brazil, marking its eighth data center in South America. This expansion aims to improve performance and security for over six million Internet applications in Brazil and provide redundancy for the São Paulo data center. The post also touches on the historical significance of Rio de Janeiro in Brazil's internet history and the city's current connectivity landscape, including community-driven ISPs in favelas.
  177. Less Is More - Why The IPv6 Switch Is MissingThis post announces the decision to make IPv6 always-on for free domains, removing the toggle and retroactively enabling it for millions of domains. It also details a collaboration with OpenDNS to prototype a new DNS metatype (TYPE65535) that returns both IPv4 and IPv6 addresses in a single query, aiming to further accelerate IPv6 adoption and performance by reducing DNS lookup round trips and improving Happy Eyeballs client behavior.
  178. Detroit and San Diego Data Centers expand Cloudflare network to 26 North American citiesThis post announces the addition of two new data centers in Detroit and San Diego, expanding the North American network to 26 cities. It highlights the strategic importance of these locations near international borders and mentions peering with local exchange points like DET-IX to improve regional interconnection and serve traffic locally.
  179. Cape Town (South Africa): Cloudflare Data Center #113This post announces the launch of Cloudflare's 113th data center in Cape Town, South Africa. This expansion brings Cloudflare 870 miles closer to millions of Internet users in the region, reinforcing its commitment to improving Internet performance and security in South Africa and Southern Africa. The new data center joins existing facilities in Johannesburg, Cairo, Luanda, Mombasa, and Djibouti, bringing Cloudflare's total African data centers to six. The post highlights the near-equivalent performance observed between Cape Town and Johannesburg after the deployment, as evidenced by latency metrics.
  180. Colombo, Sri Lanka: Six million Internet properties now faster for six million Internet usersThis post announces the addition of a new data center in Colombo, Sri Lanka, bringing the total to 112 globally and 38 in Asia. This expansion aims to improve performance for six million Internet properties and users in Sri Lanka by reducing latency, as evidenced by a 4x latency decrease reported by Cedexis.
  181. Curaçao and Djibouti - two new Cloudflare data centers located where undersea cables meetThis post announces the launch of two new Cloudflare data centers in Curaçao and Djibouti. It details the strategic importance of these locations due to their proximity to undersea cable landing points, highlighting the specific cables that connect to these regions. The post also discusses the role of these data centers in improving connectivity for the Caribbean and Eastern/Central Africa, and mentions their association with local Internet Exchange Points (IXPs).
  182. Zdravo, Beograde! Cloudflare network spans 30 European citiesThis post announces the addition of a new data center in Belgrade, Serbia, to Cloudflare's global network. This expansion increases the number of European cities in Cloudflare's network to 30, aiming to improve performance and security for websites and users in the region.
  183. Kansas City: Cloudflare’s 106th Data Center Now LiveAnnounces the launch of Cloudflare's 106th data center in Kansas City, Missouri, expanding its global network to improve performance and security for websites in the Midwest region of the United States.
  184. Buongiorno, Roma! Cloudflare Data Center CVAnnounces the launch of a new data center in Rome, Italy, bringing the total to 105. This expansion improves performance and security for users in Italy and the surrounding region by reducing latency and increasing redundancy. Rome serves as a point of redundancy to the Milan data center and expands Cloudflare's European network to 29 cities. The post also mentions the upcoming 106th data center in North America and highlights partnerships with Italian providers like Altervista, Planetel, and Aruba.it.
  185. ¡Hola, Ecuador! Quito Data Center expands Cloudflare network to 104 cities across 52 countriesThis post announces the deployment of a new data center in Quito, Ecuador, expanding Cloudflare's network to 104 cities globally and six cities in Latin America. It highlights the partnership with NAP.EC and the strategic importance of local interconnection for improving performance and reducing latency for users in the region.
  186. Panama expands Cloudflare network to 50 countriesThis post announces the launch of a new data center in Panama City, Panama, bringing Cloudflare's global network to 102 data centers and spanning 50 countries. It highlights the improvement in latency for end-users in Panama and mentions ongoing network expansion plans for 2017.
  187. Not one, not two, but three undersea cables cut in JerseyThis post demonstrates the practical application of Cloudflare's global network resilience in the face of a significant infrastructure failure. It showcases how the network automatically rerouted traffic via BGP from a severed undersea cable connection to Jersey, utilizing an alternative path from Paris instead of the usual London route. This event highlights the robustness of Cloudflare's infrastructure in maintaining service continuity during unexpected disruptions.
  188. 98.01% of sites on Cloudflare now use IPv6This post highlights the significant achievement of 98.01% of sites on Cloudflare using IPv6, demonstrating the platform's role in driving the adoption of this critical internet protocol. It details the performance benefits of IPv6, identifies key countries and networks leading adoption, and analyzes the impact of different devices, operating systems, and browsers on IPv6 usage. The post also showcases Cloudflare's proactive efforts in enabling IPv6 for its customers and its commitment to future internet technologies.
  189. The Internet is Hostile: Building a More Resilient NetworkThis post details Cloudflare's efforts to build a more resilient network by actively monitoring the internet's health using its global network infrastructure. It introduces the concept of 'internet brownouts' and explains how Cloudflare leverages BGP, custom monitoring tools, and open-source software (ZeroMQ, Prometheus, OpenTSDB) to detect and automatically mitigate network issues. The post highlights the development of an automated mitigation system, inspired by DDoS mitigation techniques, using tools like Napalm and Salt to manage network devices and reroute traffic during failures, ultimately improving performance and reducing customer errors.
  190. Amsterdam to Zhuzhou: Cloudflare network expands to 100 citiesThis post announces the expansion of Cloudflare's global network to 100 cities with the addition of 14 new data centers in Luanda, Boston, Las Vegas, Nashville, Omaha, Philadelphia, St Louis, Tampa, Athens, Lisbon, Helsinki, Brisbane, Manila, and Shanghai. The expansion aims to reduce latency, improve security, and increase redundancy for millions of websites by serving web requests closer to internet users. The post also highlights ongoing upgrades to server hardware, networking gear, and network capacity, as well as improvements in availability and automation.
  191. Evenly Distributed FutureThis post discusses Cloudflare's mission to democratize the internet's future by making advanced technologies and performance accessible to all websites, regardless of size. It highlights how Cloudflare's global network and continuous adoption of new internet technologies (IPv6, HTTP/2, TLS 1.3, Brotli) provide a performance and security advantage, effectively bridging the gap between large, well-resourced organizations and smaller websites. It also touches on how Cloudflare's infrastructure helps mitigate the impact of unexpected popularity and DDoS attacks, and contrasts the limitations of hardware-based solutions with Cloudflare's software-defined approach.
  192. Bandwidth Costs Around the WorldThis post analyzes the relative cost of bandwidth around the world, comparing transit vs. peering costs in Europe, North America, Africa, the Middle East, Asia, South America, and Oceania. It details how Cloudflare's network expansion and peering strategy influence effective bandwidth costs in different regions, highlighting specific challenges and opportunities in markets like South Korea and South America.
  193. A Post Mortem on this Morning's IncidentThis post details two incidents of significant packet loss on a major transit provider (Telia Carrier) in June 2016. It highlights the impact on Cloudflare customers (522 errors) and outlines Cloudflare's response, including taking down ports with the problematic provider and shifting traffic. Crucially, it discusses the need for improved communication during incidents and the ongoing work to build a more resilient network through greater interconnection, automated mitigation systems (augmenting BGP to detect packet loss), and increased failover capacity. The post announces plans to extend automated detection and mitigation capabilities to all POPs.
  194. Moscow, Russia: CloudFlare’s 83rd data centerThis post announces the launch of Cloudflare's 83rd data center in Moscow, Russia. It details the participation in three Internet Exchange Points (MSK-IX, Data IX, and Global IX) to increase global interconnectedness. The primary technical benefit highlighted is the reduction of latency for transcontinental traffic within Russia by over 20ms, improving performance for millions of users and applications.
  195. Hello, Colorado! CloudFlare's 82nd Data Center is Live in DenverCloudflare launched a new data center in Denver, Colorado, its 82nd globally and 12th in the United States. This expansion aims to improve regional web performance by reducing latency for users in the Denver area. The post highlights Denver's growing tech scene and the importance of regional interconnection. Cloudflare participates in two major internet exchanges in Denver: Any2 Denver and IX-Denver. The launch is expected to halve latency for end-users in Denver.
  196. Brussels, Belgium: CloudFlare's 81st Data Center is Live!This post announces the launch of a new Cloudflare data center in Brussels, Belgium. It highlights the strategic importance of Brussels for European institutions and its high internet penetration. The new data center enhances redundancy with existing facilities in Amsterdam and Paris and enables traffic exchange at the Belgium National Internet Exchange (BNIX). This expansion contributes to Cloudflare's ongoing global network growth, bringing the total to 81 data centers worldwide and 23 in Europe.
  197. Perth, Australia: 80th Data CenterThis post announces the launch of Cloudflare's 80th data center in Perth, Australia, marking its fourth in the Oceania region. It highlights the geographical proximity of Perth to Singapore and its connection to the Western Australia Internet Exchange (WA-IX). The post also includes a visual representation of latency improvements and a comprehensive list of cities where Cloudflare has data centers.
  198. Supporting the transition to IPv6-only networking services for iOSThis post details Cloudflare's proactive support for the transition to IPv6-only networking, specifically addressing Apple's requirement for iOS apps. It explains the technical reasons for the transition, the challenges involved in global IPv6 adoption, and how Cloudflare's existing IPv6 support, including an IPv6-to-IPv4 gateway and Pseudo IPv4 features, simplifies this migration for customers. It also highlights the performance and security benefits of IPv6 adoption.
  199. Bangkok, Thailand: CloudFlare’s 79th Data CenterThis post announces the launch of Cloudflare's 79th data center in Bangkok, Thailand. This expansion increases the global network to 79 cities and specifically targets the growing mobile internet usage in Thailand by leveraging ChaCha20-Poly1305 cipher suites and HTTP/2 Server Push for improved performance and battery life. The deployment also includes a partnership with JasTel Network Company to improve local internet experience for their customers. The post also mentions the Bangkok Neutral Internet Exchange (BKINX) as a potential improvement for Thailand's interconnection.
  200. Kyiv, Ukraine: Cloudflare’s 78th Data CenterThis post announces the launch of Cloudflare's 78th data center globally and 22nd in Europe, located in Kyiv, Ukraine. The expansion aims to localize content, improve performance by reducing latency for website visitors, and increase redundancy and resilience against attacks. The company will serve traffic to ISPs that peer locally, encouraging others to do the same.
  201. Taipei: CloudFlare’s 77th Data Center is Now LiveCloudflare launched its 77th data center globally and 28th in Asia, located in Taipei, Taiwan. This new data center will serve millions of websites locally, improving performance by up to 4x and significantly increasing availability for Taiwanese users by reducing latency and avoiding undersea cable cuts. The post highlights the challenges of delivering traffic to Taipei from other Asian hubs and the benefits of localized edge presence.
  202. CloudFlare launches new data centers in Oslo and MinneapolisCloudflare has launched new data centers in Oslo, Norway (#75) and Minneapolis, USA (#76). The Oslo data center is Cloudflare's third in Scandinavia and its northernmost to date. It connects to the Norwegian Internet Exchange (NIX) and FIXO. The Minneapolis data center serves the Twin Cities and the Midwest, connecting to the Midwest Internet Cooperative Exchange (MICE). Both deployments aim to improve network performance and coverage.
  203. Think Global, Peer Local. Peer with CloudFlare at 100 Internet Exchange PointsThis post details Cloudflare's expansion to its 100th Internet Exchange Point (IXP), highlighting the technical importance of peering for network performance, redundancy, and DDoS mitigation. It explains the concept of peering, the technical breakdown of IXPs as Layer 2 LANs, and the role of BGP in exchanging routes. The post emphasizes how direct peering at IXPs reduces latency by avoiding intermediary networks and improves resilience by providing diverse traffic paths. It also explains how peering helps in counteracting DDoS attacks by providing a direct path for malicious traffic to Cloudflare's filtering infrastructure.
  204. Cairo, Egypt: CloudFlare’s 74th Data CenterCloudflare launched its 74th data center in Cairo, Egypt, in partnership with Telecom Egypt. This marks their third data center in Africa and provides a 2x performance improvement for Egyptian websites. The deployment of caches directly into ISP facilities, like Telecom Egypt, is a global strategy to improve performance, reduce origin load, and provide local interconnects.
  205. Vancouver & Montreal, Canada: CloudFlare's latest data centersThis post announces the launch of two new data centers in Montreal and Vancouver, Canada, bringing Cloudflare's total points of presence (PoPs) to 70 and 71 globally. This expansion aims to reduce latency for Canadian internet users, particularly for e-commerce transactions, by bringing Cloudflare's network closer to the country's 31 million internet users. The post also references existing content on HTTP/2 optimization.
  206. Partial kernel bypass merged into netmap mainThis post details Cloudflare's contribution to the netmap project, specifically the merging of a partial kernel bypass feature into the netmap main branch. This enhancement allows for more configurable network interface modes, enabling specific queues to be detached from the host network stack while others remain attached. This improves performance by allowing direct access to network packets for applications, a significant step in optimizing network I/O for high-throughput scenarios.
  207. CloudFlare launches India data centers in Mumbai, Chennai and New DelhiCloudflare announced the launch of three new Points of Presence (PoPs) in Mumbai, Chennai, and New Delhi, India, bringing its total to 68 data centers globally. These new PoPs aim to reduce latency for Indian users to under 50ms, significantly improving performance for local traffic and applications hosted outside of India.
  208. Copenhagen, Denmark: CloudFlare's 65th Data CenterCloudflare launched its 65th data center in Copenhagen, Denmark, extending its network to 34 countries and 17 in Europe. This expansion aims to reduce median latency to under 40ms across Europe and improve performance for web applications, mobile apps, and APIs. The new data center is already mitigating DDoS attacks and contributing to the company's goal of launching one data center per week throughout 2015.
  209. Phoenix, AZ: CloudFlare's 64th data centerThis post announces the launch of Cloudflare's 64th data center in Phoenix, AZ. It details the importance of regionalizing traffic to improve performance and reduce latency, especially for users outside major interconnection hubs. The post explains the concept of 'hot potato routing' and uses measurements of BGP community tags from Tier 1 providers (NTT, TeliaSonera, Tata Communications, Cogent) to illustrate the low level of local interconnection in Phoenix compared to the San Francisco Bay Area. The launch of the Phoenix data center is presented as a solution to bring content closer to local audiences.
  210. Manchester, UK: CloudFlare's 63rd data centerAnnounced the launch of a new data center in Manchester, UK, bringing the total to 63 PoPs in 33 countries. This expansion improves redundancy and content localization within the UK, serving Northern England with reduced latency. Cloudflare also joined IX Manchester, a peering exchange.
  211. iOS 9 — How did the launch really go?This post provides insights into the adoption rates of iOS 9 by analyzing user agent strings passing through Cloudflare's network. It highlights the impact of the iOS 9 launch on IPv6 adoption, demonstrating Cloudflare's ability to leverage its global network visibility to track significant internet events and the adoption of new technologies.
  212. How We Extended CloudFlare's Performance and Security Into Mainland ChinaThis post details the technical, economic, and regulatory challenges of extending Cloudflare's performance and security platform into mainland China. It describes the partnership with Baidu to overcome these challenges, including Baidu's expertise in ICP licensing and routing within China's ISP landscape. The post also announces the launch of 17 data centers in China and the co-launch of Baidu's Yunjiasu service, highlighting significant performance improvements for Chinese internet users.
  213. Kuala Lumpur, Malaysia: CloudFlare's 45th data centerThis post announces the deployment of Cloudflare's 45th global point of presence (PoP) in Kuala Lumpur, Malaysia. It highlights the technical challenge of 'boomeranging' traffic due to limited domestic interconnection in certain Asian regions and explains how Cloudflare's membership in the MyIX neutral Internet Exchange Point (IXP) resolves this issue, enabling local interconnection of ISPs. This improves load times and application performance for Malaysian users and Cloudflare customers.
  214. Berlin, Germany: CloudFlare's 44th Data CenterCloudflare has launched a new data center in Berlin, Germany, marking its 44th globally and 14th in Europe. This expansion is part of a broader strategy to increase Points of Presence (PoPs) in major European cities to improve network performance and resilience.
  215. Test all the things: IPv6, HTTP/2, SHA-2This post details the launch of a test server to allow users to test compatibility with IPv6, HTTP/2, and SHA-2 certificates. It showcases Cloudflare's proactive approach to adopting and enabling new internet protocols and security standards, specifically highlighting the dual-stack IPv4/IPv6 support and the use of HTTP/2 for faster web delivery. The post also emphasizes the importance of SHA-2 over the deprecated SHA-1 for enhanced security.
  216. Mombasa, Kenya: CloudFlare's 43rd data centerThis post announces the deployment of a new Cloudflare Point of Presence (PoP) in Mombasa, Kenya. This marks Cloudflare's 43rd global PoP and second in Africa. The primary technical contribution is the establishment of local infrastructure to serve traffic to East African countries, significantly reducing latency for users in Kenya, Tanzania, Rwanda, Uganda, Ethiopia, and Mauritius. Latency improvements are quantified with specific examples, showing a 7.5x reduction in round trip time for Uganda and Tanzania. The post also contextualizes this expansion within the growth of Internet infrastructure in Africa and the benefits of localized content delivery.
  217. Now serving the Middle East: 4 new data centers, partnershipsThis post announces the deployment of four new data centers in the Middle East (Doha, Dubai, Kuwait City, Muscat), bringing the total to 42. It details the rationale behind this expansion, focusing on the high cost of local bandwidth in the MENA region and the suboptimal international connectivity for regional traffic. The post highlights partnerships with regional carriers to improve infrastructure and reduce latency for Cloudflare users in the region, citing significant latency reductions in the UAE and Qatar.
  218. Célébrer le 14 Juillet avec Marseille, le 36ème point de présence de CloudFlareThis post announces the addition of a new data center in Marseille, France, bringing the global total to 36. It highlights Marseille's strategic importance as a gateway for Mediterranean traffic and its role in reducing latency for users in Algeria and other countries. The deployment also includes participation in the FranceIX Marseille Internet exchange.
  219. Blue Light Special: Ensuring fast global configuration changesThis post details the challenge of propagating configuration changes across Cloudflare's global network in near real-time, specifically for DNS records and WAF rules. It introduces a custom alert system using Raspberry Pis and a visible police light to monitor propagation times from an external perspective, ensuring timely awareness of potential issues.
  220. Osaka, Japan: CloudFlare's 35th Data CenterThis post announces the deployment of Cloudflare's 35th global data center in Osaka, Japan. It details the strategic importance of this location for serving the Keihanshin metropolitan area and improving application delivery to 30 million inhabitants. The post also discusses the challenges of internet traffic delivery in Asia due to poor routing and lack of local interconnection among incumbent telecommunications providers, contrasting it with the more developed IXP ecosystem in Europe. Cloudflare's membership in multiple Japanese IXPs (JPIX, JPNAP, BBIX, Equinix) is highlighted as a means to leverage its Anycast network architecture for improved routing and reduced latency for users in Nagoya and the surrounding region.
  221. iOS Developers — Migrate to iOS 9 with CloudFlareThis post highlights how Cloudflare's existing support for IPv6 and HTTPS, particularly its automatic IPv6 gateway and Universal SSL, directly addresses new requirements imposed by Apple for iOS 9 and OS X El Capitan. It emphasizes that developers can leverage Cloudflare to meet these mandates without complex backend changes, ensuring their applications work globally and securely. The post also reinforces Cloudflare's role in making modern network protocols accessible and affordable.
  222. Four years later and CloudFlare is still doing IPv6 automaticallyThis post highlights Cloudflare's four-year commitment to automatic IPv6 support, demonstrating its significant role in enabling IPv6 connectivity for over two million websites. It showcases the growth of IPv6 adoption through data on unique IP addresses, end-user traffic to mobile devices, and country-specific IPv6 delivery percentages, with a special mention of the Peru datacenter's high IPv6 traffic. The post reinforces Cloudflare's position as a leading global CDN for IPv4/IPv6 content delivery by default and at scale, and emphasizes the increasing importance of IPv6 due to IPv4 address depletion.
  223. Oceania Redundancy: Auckland and Melbourne data centers now onlineThis post details the deployment of new data centers in Auckland and Melbourne to provide regional redundancy for Oceania. It explains the rationale behind choosing these locations over a second Sydney data center, focusing on network topology, latency reduction, and peering opportunities. The challenges of fiber infrastructure in Auckland are also discussed.
  224. CloudFlare's Buenos Aires data center now onlineAnnounces the deployment of Cloudflare's 32nd data center in Buenos Aires, Argentina. This deployment is the 5th in Latin America and aims to reduce latency for Internet users in Argentina by enabling local content caching and traffic exchange with ISPs through participation in the CABASE IXP. This reduces reliance on distant data centers in Newark and Miami.
  225. CloudFlare in 2014: Bigger, Faster, SecurerThis post details the expansion of Cloudflare's global network in 2014, with the addition of data centers in Chile, Spain, Italy, Brazil, Colombia, Peru, and South Africa. It also highlights the rollout of new protocols and security features.
  226. Johannesburg: CloudFlare’s 30th data centerThis post announces the deployment of Cloudflare's 30th data center in Johannesburg, South Africa, marking the company's first presence on the African continent. This expansion significantly reduces latency for Internet users in South Africa and the broader Southern African region, with latency improvements ranging from 2x to 100x depending on the location. The deployment was made possible through partnerships with regional providers and addresses the high cost of bandwidth in Africa.
  227. Lima, Peru: CloudFlare’s 29th data centerThis post announces the launch of Cloudflare's 29th data center in Lima, Peru, marking a significant expansion into Latin America. The new data center aims to reduce latency for users in the region, increase web performance, and add redundancy to the network. It also enhances Cloudflare's capacity to absorb cyber attacks, benefiting local customers like the Presidency of Peru and ONPE. The company also mentions ongoing efforts to deploy over 20 new data center sites globally.
  228. Route leak incident on October 2, 2014This post details a specific incident of a BGP route leak from an ISP (Internexa) that caused significant downtime for Cloudflare customers. It highlights the fragility of internet routing protocols and the impact of such leaks on global network availability. Cloudflare's response involved direct collaboration with the ISP, quarantining a data center, and disabling connectivity with the problematic provider, demonstrating their operational response to network disruptions. The post also emphasizes the ongoing challenge of route leakage and Cloudflare's commitment to service availability and post-mortem analysis.
  229. One More Thing: Keyless SSL and CloudFlare's Growing NetworkThis post introduces Keyless SSL as a critical technology enabling Cloudflare's continued global network expansion. It explains how Keyless SSL removes the constraint of storing private SSL keys on edge servers, allowing Cloudflare to deploy infrastructure in a wider range of locations, including those with less stringent physical security or where geopolitical concerns about data storage exist. This directly supports the goal of a larger, more performant, and more resilient network.
  230. The Relative Cost of Bandwidth Around the WorldThis post provides a detailed analysis of the relative costs of bandwidth around the world, explaining the concepts of transit and peering. It breaks down these costs by region (North America, Europe, Asia), using illustrative benchmarks and data on Cloudflare's peering vs. transit traffic split. The post explains how peering agreements, especially those facilitated by non-profit exchanges in Europe, significantly reduce effective bandwidth costs compared to regions with more expensive transit and for-profit exchanges.
  231. Listo! Medellin, Colombia: CloudFlare's 28th Data CenterThis post announces the deployment of Cloudflare's 28th data center in Medellin, Colombia, in partnership with Internexa. This expansion localizes content access within Latin America, reducing latency for the region's internet users and making all sites using Cloudflare available locally over Internexa's IP backbone.
  232. CloudFlare Joins Three More Peering Exchanges in AustraliaThis post details Cloudflare's establishment of connections to three new Internet exchanges in Australia (Megaport in Sydney, Brisbane, and Melbourne), doubling its presence in Australia and marking its first participation outside of Sydney. It explains the benefits of peering for reducing latency and improving performance by exchanging traffic directly with ISPs, rather than through third-party transit providers. The post also mentions ongoing global network expansion and participation in other upcoming exchanges.
  233. Three years after World IPv6 DayThis post highlights Cloudflare's significant role in the adoption of IPv6, noting that 20% of the IPv6 web is accessible via Cloudflare customers. It details the challenges faced by some websites in enabling IPv6 due to legacy systems and introduces a 'pseudo IPv4' solution as a temporary workaround to facilitate wider adoption. The post emphasizes Cloudflare's commitment to encouraging full IPv6 enablement for its users.
  234. I joined CloudFlare on Monday along with 5,000 othersThis post marks the author's first day as a Network Strategy employee at Cloudflare and highlights the company's significant growth, evidenced by 5,000 new customer sign-ups on the same day. A major contribution is the announcement and detailed explanation of Cloudflare's strategic shift to make IPv6 the default setting for new customers and to gradually enable it for existing customers. This initiative aims to accelerate the adoption of IPv6, addressing the scarcity of IPv4 addresses and ensuring the continued growth of the internet.
  235. Stories from our recent global data center upgradeThis post details the significant effort and logistical challenges involved in doubling the capacity of Cloudflare's global network over six months. It highlights the team's dedication and problem-solving skills in overcoming surprises during data center upgrades across multiple international locations, including specific anecdotes about hardware procurement, installation, and overcoming customs delays. The post also outlines future expansion plans for Latin America, the Middle East, Africa, and Asia.
  236. Bienvenido a Chile: Cloudflare's 24th Data Center Now Live!This post announces the launch of Cloudflare's 24th data center in Valparaiso, Chile, marking its first presence in Latin America. This expansion aims to reduce latency for users in Chile and Argentina by bringing traffic closer to them, rather than routing through Miami. The post highlights the logistical complexities of launching a new data center and hints at future expansions in Latin America.
  237. 2013: Rebuild the Engine; 2014: Step on the GasThis post details Cloudflare's infrastructure upgrades in 2013, including a move to 10Gbps per-server infrastructure, a complete rebuild of the DNS infrastructure in Go, standardization of the HTTP request pipeline around Lua, and the creation of a new rules-based WAF. It also outlines plans for significant global network expansion in 2014, including new data centers in Latin America, the Middle East, Africa, and Asia, with a focus on China and Brazil.
  238. A Tour Inside CloudFlare's Latest Generation ServersThis post details the evolution of Cloudflare's server hardware, specifically the G4 generation, highlighting upgrades in storage (SSDs, increased RAM for caching), CPU (low-power Sandybridge for efficiency and future features), and network interfaces (10Gbps Solarflare cards with kernel bypass technology). This hardware evolution is presented as a direct enabler of their global network expansion and ability to handle increasing traffic and new service demands.
  239. Happy IPv6 Day: Usage On the Rise, Attacks TooThis post highlights Cloudflare's proactive support for IPv6, detailing its growth on the network, its role as a major IPv6 provider through a free gateway, and the increasing adoption by mobile networks. It also identifies a new trend of IPv6-based attacks, primarily DDoS, and discusses the challenges and opportunities IPv6 presents for threat management.
  240. Today's Network IssueThis post details a network issue caused by an upstream transit provider's network consolidation and migration, which inadvertently rerouted traffic to Cloudflare's Los Angeles data center, causing overload and service degradation for a portion of users. It highlights Cloudflare's reliance on Anycast routing, the challenges of managing traffic flow through multiple transit providers, and the company's rapid response and collaboration with the transit provider to resolve the issue. The post also emphasizes Cloudflare's commitment to learning from such incidents and improving its internal routing management to prevent future disruptions.
  241. Syrian Internet RestoredThis post details Cloudflare's analysis of the Syrian internet outage and subsequent restoration. It showcases Cloudflare's capability to monitor BGP route withdrawals and re-announcements across national networks, providing evidence against the government's claim of a cable cut and instead pointing to a systematic shutdown. The post also highlights the observation of a small portion of Syrian IP space that maintained connectivity, further discrediting the official explanation and demonstrating Cloudflare's insight into network behavior during critical events.
  242. How Syria Turned Off the Internet (Again)This post details Cloudflare's real-time monitoring and analysis of BGP route withdrawals, specifically in response to Syria's internet shutdown. It highlights Cloudflare's capability to detect and report on such events, showcasing its role in observing and understanding global internet connectivity disruptions.
  243. How the CloudFlare Team Got Into Bondage (It's Not What You Think)This post details Cloudflare's implementation of 'Project Bondage,' which involves port bonding (specifically 802.3ad Dynamic Link Aggregation) and custom IRQ handling. These technical optimizations significantly increase network throughput per server and improve resilience against SYN flood attacks by isolating traffic and preventing bottlenecks, thereby enhancing the overall performance and stability of the global network.
  244. Load Balancing without Load BalancersThis post details Cloudflare's innovative use of Anycast routing at both the Wide Area Network (WAN) and Local Area Network (LAN) levels to achieve load balancing and fault tolerance without relying on traditional hardware load balancers. It explains how Anycast directs traffic to the nearest available data center and how, within a data center, BGP routes are managed by software like Bird to distribute traffic across servers and automatically reroute around failures. The post also touches on early experiments with BGP for more sophisticated load balancing, particularly for protocols like UDP.
  245. Today's Outage Post MortemThis post details a significant outage caused by a bug in a Flowspec rule applied to Juniper edge routers, leading to a system-wide failure and temporary internet disconnection. It highlights the importance of BGP and Flowspec in network management, the challenges of automated rule propagation, and the incident response process to restore services. The post also draws parallels to the Syrian internet outage, emphasizing the critical nature of routing protocols.
  246. Today's System-Wide UpgradeThis post details a system-wide migration of over half a million websites to a new provisioning system. The migration involved rebalancing sites to new IP address pools, seamlessly issuing and redeploying SSL certificates (including custom ones), and addressing two specific issues: a staging SSL environment conflict causing temporary SSL errors for a handful of customers, and a brief period of unreachability for a small number of sites due to an incorrectly recorded null route on an IP address that had been under DDoS attack. The new system provides significantly more flexibility and robustness, enabling features like dynamic site tagging for content restrictions and attack mitigation, and improved isolation of attacked sites.
  247. Syrian Internet access reestablished starting 1432 UTCThis post details the reestablishment of internet connectivity in Syria following an outage. It uses BGP maps and internal monitoring data to show the phased restoration of connections with global carriers (PCCW, TATA, Turk Telecom) and the return of traffic levels to pre-outage levels. The analysis highlights Cloudflare's role in observing and reporting on the restoration of internet access.
  248. How Syria Turned Off the InternetThis post details the technical observation and analysis of the Syrian internet outage on November 29, 2012. It explains how BGP routes to Syrian IP space (AS29386) were simultaneously withdrawn from all upstream providers (PCCW, Turk Telekom, Telecom Italia, TATA), effectively disconnecting the country. The analysis suggests this was a deliberate configuration update rather than a physical cable cut, based on the systematic withdrawal of routes and the fact that all four physical connections would have needed to be cut simultaneously. A video created with BGPlay visualizes the route withdrawal process.
  249. Why Google Went Offline Today and a Bit about How the Internet WorksThis post details Cloudflare's role in diagnosing and helping to resolve a significant Google outage caused by a BGP route leak originating from an Indonesian ISP (Moratel). It highlights the importance of Cloudflare's network engineering team's around-the-clock monitoring of internet routing (BGP) and their ability to quickly identify and communicate with other network operators to fix such issues. The post also serves as a case study illustrating the inherent trust-based vulnerabilities in BGP and how Cloudflare's global network and expertise are crucial for maintaining internet stability and ensuring optimal routes for its customers, even when external factors cause widespread disruptions.
  250. What Happens When a Hurricane Hits the WebThis post details Cloudflare's infrastructure resilience during Hurricane Sandy, including the redundancy of its data centers (EWR, IAD), backup power systems (batteries, generators), and the network's ability to reroute traffic around failures like the AC-2 undersea cable incident. It also discusses the 'Always Online' functionality serving static versions of sites when origin servers were down and observations on traffic patterns during the storm.
  251. Seoul, Korea: CloudFlare's 23rd Data CenterThis post announces the launch of Cloudflare's 23rd data center in Seoul, Korea, marking the final city in a recent expansion wave. It highlights Seoul's advanced internet infrastructure and the challenges of trans-oceanic cable bottlenecks. The new data center will improve end-to-end performance for Korean users and enhance DDoS mitigation capabilities against local botnets.
  252. Warsaw, Poland: CloudFlare's 22nd Data CenterThis post announces the launch of Cloudflare's 22nd data center in Warsaw, Poland, which will improve performance in Northeastern Europe and Russia, and reduce load on the Frankfurt data center. It also mentions plans to double the size of the London facility and potential future European locations (Moscow, Milan, Madrid, Sofia, Istanbul).
  253. Stockholm, Sweden: CloudFlare's 21st Data CenterThis post announces the launch of Cloudflare's 21st data center in Stockholm, Sweden. It details the historical expansion of Cloudflare's data center footprint, listing previous locations and their opening order. The post highlights the strategic importance of Stockholm as a hub for the Nordic countries and its expected impact on reducing ping times in Northern Europe. It also mentions Cloudflare's partnership with Pingdom, a Swedish company, and the author's personal excitement about pressing the key to activate the new data center.
  254. Prague, Czech Republic: CloudFlare's 20th Data CenterThis post announces the launch of Cloudflare's 20th data center in Prague, Czech Republic, and the addition of a new network provider for inbound traffic. It highlights the importance of Anycast routing for performance and DDoS mitigation, and details the ongoing work by the network operations team to ensure network stability with continued expansion.
  255. Toronto, Canada: CloudFlare's 18th Data CenterThis post announces the launch of Cloudflare's 18th data center in Toronto, Canada. This new facility will improve performance for Canadian users and offload traffic from existing data centers in Chicago and Newark. The post also mentions plans to launch five more facilities in Europe and Asia in the coming days.
  256. CloudFlare (aka. KickassDNS)This post highlights Cloudflare's development of one of the largest, fastest, and most resilient authoritative DNS networks in the world. It details how the use of Anycast technology, where every server in every data center can answer any DNS query, contributes to speed and fault tolerance. The post also introduces the 'KickassDNS' control panel, emphasizing its ease of use and rapid propagation of updates, and mentions the introduction of custom DNS servers for Business and Enterprise plans, as well as partnerships with hosting providers.
  257. Seattle: CloudFlare's 17th Data CenterAnnounces the launch of Cloudflare's 17th data center in Seattle, Washington. This new facility will serve traffic from the Pacific Northwest and offload load from the busy San Jose data center. The post also mentions crossing 400 billion page views and 40 petabytes of bandwidth saved, and notes a change in the network map's color scheme from blue land to shades of gray.
  258. Sydney, Australia: CloudFlare's 15th Data CenterThis post announces the launch of Cloudflare's 15th data center in Sydney, Australia. This expansion aims to improve network performance for users in Australia and New Zealand by reducing ping times from over 160ms to under 40ms. The post details the challenges of securing bandwidth providers in the region and the eventual success in establishing a new facility.
  259. The bandwidth of a Boeing 747 and its impact on web browsingThis post introduces the concepts of bandwidth and latency as key factors affecting web browsing speed. It explains how TCP's reliability mechanisms, particularly acknowledgements, are limited by latency and the speed of light, impacting download speeds. The post then positions Cloudflare's distributed network of data centers as a solution to this problem by reducing the physical distance between users and their website's content, effectively lowering latency.
  260. Global IPv6 Challenge: No More Excuses, Enable the FutureThis post marks a significant push for IPv6 adoption among Cloudflare users and the broader internet. It introduces the 'Global IPv6 Challenge' to encourage websites to enable IPv6 support, offering a free and easy-to-use Automatic IPv6 Gateway. The post details two modes of IPv6 support (SAFE and FULL) and announces the upcoming IPv6 Network Testing app to gather data on ISP support and identify network issues. This initiative directly contributes to Cloudflare's mission of building a better web by addressing the IPv4 address space limitations and promoting the transition to IPv6.
  261. Today's Outage Post MortemThis post details a significant network outage caused by a Border Gateway Protocol (BGP) routing error during scheduled maintenance in Hong Kong. It explains the technical cause of the outage, the impact on traffic, and the immediate steps taken to resolve it. Crucially, it outlines future prevention measures, including implementing verification layers for routing changes and enhancing communication with upstream providers to prevent similar incidents. This post highlights the operational challenges and ongoing efforts to ensure the reliability and resilience of Cloudflare's global network.
  262. 30% More Traffic in Less Than a Blink of an EyeThis post details how routing improvements in the network team saved approximately 25ms of latency for requests originating from Southeast Asia (Thailand, Malaysia, Indonesia). This reduction in latency resulted in a measured 30% increase in traffic (bandwidth, hits, and page views) from these regions, demonstrating the amplified impact of performance benefits on user engagement and internet use, as explained by TCP's nature.
  263. Non-Latin/UTF8 International Domains (IDNs) Now Fully SupportedThis post announces the full support for International Domain Names (IDNs) and UTF8 characters within Cloudflare's interface. Previously, users had to convert IDNs to Punycode, an 'ugly hack'. This update allows direct use of native characters for domains and across the UI, making Cloudflare more accessible to a global audience and reflecting its international expansion efforts.
  264. Welcome DreamHost!This post announces a partnership with DreamHost, making Cloudflare's services available to all DreamHost customers through a one-click integration. It also introduces 'Cloudflare Plus,' a custom plan offered by DreamHost with bundled popular Cloudflare features, demonstrating a new model for distributing Cloudflare's capabilities through hosting partners.
  265. CloudFlare's Miami Data Center Now OnlineAnnounces the launch of Cloudflare's 14th data center in Miami, which will serve Central and South America, significantly decreasing latency and expanding capacity for the region. This new facility is expected to handle over 1.1 Gbps of traffic shortly after going online and will offload traffic from existing busy data centers in Ashburn, Dallas, and Los Angeles. Plans for further network expansion in 2012 are also mentioned.
  266. CloudFlare Expanding the IPv6 WebThis post highlights Cloudflare's role in expanding the IPv6 web. It notes that while IPv6 adoption by the top 1 million websites is still low, a significant portion of those that do support IPv6 do so through Cloudflare's Automatic IPv6 Gateway. This demonstrates Cloudflare's proactive contribution to the adoption of newer internet protocols as part of its broader mission to improve the internet.
  267. 100 Billion Page ViewsThis post announces that Cloudflare will cross 100 billion page views in the last year, with 25 billion in the last 30 days. The network is averaging over 60,000 requests per second, with peak loads higher. Cloudflare ranks among the top websites globally for page views and unique visitors. The company has over 150 Gbps of capacity, plans to launch its 14th data center by year-end, and is increasing server capacity by 40% in 9 existing data centers.
  268. Groovy Baby! CloudFlare's London Data Center Now OnlineThis post announces the launch of Cloudflare's London (LHR) data center, marking its 13th global data center and fourth in Europe. The new data center is already handling over 1Gbps of traffic, diverting it from the Amsterdam data center. This expansion is driven by user and partner requests and aligns with Cloudflare's goal of making the internet fast and safe globally. Another data center is planned for rollout in approximately three weeks.
  269. Wow, That's Fast! Instant DNS Updates and More...This post introduces KTFS, a new key-value based file system designed for rapid, scalable deployment of changes across Cloudflare's network. It highlights KTFS's role in speeding up SSL certificate provisioning and, significantly, in reducing DNS update propagation time from five minutes to under a minute. This demonstrates an internal infrastructure development directly supporting the speed and efficiency of services delivered over the global network.
  270. A Brief Primer on AnycastThis post introduces and explains the Anycast routing scheme, contrasting it with Unicast. It details how Anycast enables multiple machines to share a single IP address, with routers directing traffic to the closest machine. The benefits of Anycast for Cloudflare are highlighted: speed (reduced latency by serving requests from the edge), resiliency (traffic automatically reroutes around data center outages), and attack mitigation (distributed nature absorbs DDoS attacks). The post also touches on the complexity of implementing Anycast, requiring ownership of hardware, direct carrier relationships, and route tuning.
  271. CloudFlare's Singapore Data Center Now OnlineCloudflare announced the launch of its 12th data center in Singapore, expanding its global network to protect and improve performance for users in the region. This addition brings the total number of data centers to 12, with plans to add London and Miami in the coming months, aiming to be within a few hops of virtually every internet user.
  272. 2,000 Page Views Per SecondThis post announces Cloudflare's achievement of delivering a sustained average of 2,000 page views per second, with peak traffic being even higher. It highlights the scale of log generation (over a million log lines per minute) and the shift from manual log monitoring to a global traffic display. This demonstrates the company's ability to scale its network infrastructure to handle significant traffic loads.
  273. People Just Liked It Better That Way!Cloudflare has upgraded its Amsterdam data center with custom-built hardware optimized for network-heavy loads, increasing its capacity by at least 10x. New data centers are being rolled out in Paris and Frankfurt, with London planned next, covering major Western European internet exchange points.
  274. New IPs + New Year = Big Stuff to Come!Cloudflare announced the issuance of a new IP address range (173.245.48.0/20) which will be deployed soon. This new range, along with existing ranges, will be used by Cloudflare's public-facing IPs. The company advised users filtering connections based on IP ranges to upgrade their configurations, providing instructions for mod_cloudflare. This expansion is described as the first step before a dramatic expansion of the Cloudflare network in the new year.