Problem

PgDog addresses PostgreSQL’s scaling ceiling. PostgreSQL is process-based and single-primary, which creates hard limits on client connections, query throughput, and write capacity. When a fast-growing application hits that wall, teams may have to batch requests, defer workloads, vertically scale expensive database hardware, or spend years and roughly ten engineers building sharding infrastructure themselves. The killer use case is a high-throughput, large or multi-tenant Postgres deployment—such as an application serving hundreds of thousands of transactions or orders per minute—that needs more capacity without abandoning Postgres or rewriting its application.

The economics are primarily avoided engineering cost, operational risk, and forced migration to another database. PgDog’s stated goal is to make workloads involving 100 TB-plus tables and roughly one million queries per second feasible while retaining the familiar relational Postgres interface.

Product / Service

PgDog is an open-source Rust proxy that combines transaction and session connection pooling, PostgreSQL-aware load balancing, failover handling, and distributed sharding. It multiplexes thousands of application clients over a much smaller number of database connections, parses queries to distinguish reads from writes, routes reads across replicas, and directs sharded queries to the appropriate databases. It can also coordinate cross-shard queries and atomic writes, while using logical replication to move data and reshard with limited disruption.

The delivery model is deliberately deployment-agnostic: customers can run the executable through Docker or Kubernetes in RDS, Aurora, their own cloud, on-premises, or locally, typically by changing the database endpoint rather than modifying the application. The free AGPL-licensed edition is paired with a closed-source Enterprise edition that adds a control plane, real-time operational visibility, query insights, and dedicated support; Enterprise can be self-managed on-premises or in a customer’s cloud account, or fully managed by PgDog.

Market

PgDog competes in the PostgreSQL infrastructure market across connection poolers, database proxies, read/write load balancers, and distributed SQL or sharding layers. Its most direct open-source comparisons are PgBouncer and PgCat: PgBouncer is the established connection-pooling incumbent, while PgCat overlaps on multithreading and sharding. PgDog’s differentiation is combining pooling, PostgreSQL-aware routing, failover, and application-transparent sharding in one proxy; its own benchmark reports an average performance advantage over PgBouncer and stronger results than PgCat above roughly 50 connections.

The project is commercially backed but still early. As of June 2026, PgDog reported production use across dozens of deployments, more than two million queries per second, over 20 TB sharded, and more than 1.4 million Docker pulls; named users include Reducto, Ramp, Modal, Huntress, and TripStack. It also announced a $5.5 million seed round from Basis Set, YC, Pioneer Fund, and others, and was building an SLA-backed Enterprise business. No revenue figure is disclosed in the available material, so the clearest traction signals are production usage, open-source adoption, and the transition toward paid licensing and managed support.