TechPrufer
Get started
PlanetScale
PlanetScale logoPlanetScale · Feature trail

Global Read-Only Regions

2 posts · May 2022 – Oct 2026

What this trail covers

PlanetScale Portals allows users to create read-only regions for their databases in various global locations. This feature aims to reduce read latency for globally distributed applications by placing data closer to users. Users can add multiple read-only regions to their existing databases, connecting to them via specific connection strings. The system supports automatic role switching in frameworks like Ruby on Rails to direct reads to the nearest replica and writes to the primary, while also i Dedicated read replicas are additional Postgres instances that replicate asynchronously from the primary cluster. This allows for reducing read latency for a global set of clients by standing up replicas in different data centers. Users can also create multiple dedicated read replicas within the same region to isolate specific workloads like vector search or analytics from the primary OLTP traffic. Each replica can be individually sized and configured with specific RAM, latency, and performance characteristics, including instance size and disk characteristics.

All PlanetScale trails
  1. Introducing Dedicated Read Replicas for PostgresThis post introduces dedicated read replicas for Postgres databases, allowing users to create, size, and connect to additional Postgres instances that replicate asynchronously from the primary cluster. This feature addresses use cases such as reducing read latency for globally distributed clients by placing replicas in different regions, and isolating specific workloads (e.g., vector search, analytics) within the same region by configuring individual replicas with distinct instance sizes and disk characteristics.
  2. Introducing global replica credentialsIntroduces global replica credentials that automatically route reads to the nearest read replica globally, eliminating the need for code changes when adding or removing read-only regions. This leverages the PlanetScale Global Network, a CDN-like connectivity layer that terminates MySQL connections at the edge, performs connection pooling, and routes to the database. The post details the architecture of the Global Network, including its edge termination, connection pooling, and routing mechanisms. It also explains the components of a credential: Route (mapping username to clusters), Credential (authentication and tablet type), and Endpoint (hostname). The routing logic sorts clusters by latency and selects the nearest one, with failover to the next option if a region is unavailable. This enhances the utilization of read-only regions and simplifies global read routing.
  3. Introducing PlanetScale Portals: Read-only regionsThis post introduces PlanetScale Portals, a new capability that allows users to create read-only regions for their databases. It explains the benefits of co-locating data with globally distributed applications to reduce read latency, providing an example of latency reduction from ~90ms to ~3ms. It details how to connect to these read-only regions using specific connection strings and provides a comprehensive Ruby on Rails example demonstrating `database.yml` configuration, automatic role switching, and handling replication lag. Pricing details for storage and row reads are also outlined.