Throughput autoscaling: Dynamic sizing for Facebook.com
Sep 14, 2020Daniel Boeve, Kiryong Ha, Anca Agape
Summary
This post details the implementation of throughput autoscaling for Facebook.com's web tier. It describes the transition from static capacity provisioning to dynamic sizing based on workload-driven throughput metrics. The system models predicted steady-state and disaster demand using ML and estimates supply through internal load testing. This approach optimizes resource utilization by dynamically adjusting capacity to meet fluctuating global traffic patterns and disaster readiness requirements.
More from the Scalability and Large-Scale Systems Engineering trail
- Open-Sourcing Rebalancer: A Generic, High-Performance Library for Solving Assignment Problems
- ZGateway: Learnings from Putting a Proxy in Front of ZippyDB
- MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines
- From User Sequences to Scaling Laws: A Multi-Stage Architecture for Meta’s Ads Ranking
- GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model