PgDog
PgDog is a PostgreSQL proxy tool designed to simplify database scaling without requiring application changes or database extensions.
Profile
A proxy that scales PostgreSQL through connection pooling, load balancing, and sharding without requiring database extensions.
PgDog is a PostgreSQL proxy tool designed to simplify database scaling without requiring application changes or database extensions. Founded by Lev, who previously created PgCat, PgDog emerged from the need to address Postgres scaling challenges more flexibly. The tool is written in Rust and operates as an external proxy, making it compatible with managed Postgres services like RDS and Cloud SQL where extensions like Citus aren't available.
PgDog combines connection pooling, load balancing, and sharding into a single executable, deployable anywhere. It supports cross-shard transactions, replicated tables, and consistent schema changes, maintaining ACID compliance while scaling. The project gained attention with its May 2025 open-source release, highlighted by a Hacker News discussion where Lev explained its architecture and goals.
PgDog targets OLTP workloads, differentiating itself from Citus by focusing on proxy-based sharding rather than in-database coordination. Early adopters include companies like Coinbase and Ramp, though PgDog remains in active development with features like advanced aggregate functions still underway. The tool's ability to handle thousands of concurrent connections and its drop-in compatibility with existing Postgres clients position it as a pragmatic solution for teams facing scaling limits.
Who buys this
- Engineering teams hitting Postgres performance limits
- Companies using managed Postgres services without extension support
- Startups needing horizontal scaling without app rewrites
- Teams prioritizing OLTP workloads over OLAP
- Developers seeking drop-in replacements for existing Postgres connections
Publicly disclosed clients
- Coinbase
- Ramp
- Bitstack
- Reducto
- Circleback
- TripStack
- Span
Strengths and what to watch
Strengths
- Drop-in compatibility with existing Postgres clients and drivers
- Multi-threaded Rust architecture handles thousands of concurrent connections
- Works with managed Postgres services where extensions are unavailable
Watch for
- Early-stage project with incomplete features like advanced aggregates
- In-memory sorting may limit very large result sets
- Competing with mature solutions like Citus despite different architectures
Key Information
- Founded
- 2025
- Headquarters
- San Francisco
Frequently Asked Questions
What is PgDog and how does it work with PostgreSQL?
PgDog is a Rust-based PostgreSQL proxy that scales databases through connection pooling, load balancing, and sharding without requiring extensions. It operates externally, making it compatible with managed services like RDS. The tool maintains ACID compliance while handling thousands of connections as a single executable.
Can PgDog work with Amazon RDS or Google Cloud SQL?
Yes, PgDog works with managed Postgres services like RDS and Cloud SQL where extensions aren't supported. Its proxy architecture sits between applications and databases, providing scaling features without requiring database modifications. This makes it ideal for teams locked into managed service constraints.
How does PgDog compare to Citus for Postgres scaling?
PgDog differs from Citus by using proxy-based sharding rather than in-database coordination. While Citus requires extensions, PgDog operates externally, focusing on OLTP workloads. It offers drop-in compatibility with existing clients but lacks some mature features like advanced aggregates available in Citus.
What companies are using PgDog for database scaling?
Notable PgDog adopters include Coinbase, Ramp, and Bitstack, primarily engineering teams hitting Postgres performance limits. The tool appeals to startups needing horizontal scaling without app rewrites and companies using managed Postgres services where traditional extensions aren't available.
Does PgDog support cross-shard transactions in PostgreSQL?
Yes, PgDog supports cross-shard transactions while maintaining ACID compliance. It handles schema changes consistently across shards and includes replicated table support. The proxy coordinates these operations externally, unlike extension-based solutions that modify database internals.
What are the current limitations of PgDog?
As an early-stage project, PgDog lacks some features like advanced aggregate functions. In-memory sorting may limit very large result sets. While competing with mature solutions, its proxy architecture offers unique advantages for specific use cases, particularly with managed services where extensions aren't an option.
Sources
- pgdog.dev — Product capabilities and customer logos
- news.ycombinator.com — Founder background and technical approach
- techcrunch.com — Industry context for 2025 tech environment
- www.informationweek.com — Broader tech industry trends affecting database tools