Scaling Postgres Without the NoSQL Pivot: PgDog Secures $5.5M to Build a Better Database Proxy

Every developer reaches a point in an application's lifecycle where the database starts to sweat. Usually, that database is PostgreSQL. It’s reliable, feature-rich, and the default choice for almost every modern web stack. But when tables cross the multi-terabyte threshold and query volume spikes, the traditional path has been painful: either embark on a grueling manual sharding project or reluctantly migrate to a NoSQL alternative like MongoDB or DynamoDB.

A three-person startup called PgDog wants to eliminate that dilemma entirely. Backed by a fresh $5.5 million in funding from Basis Set, YC, Pioneer Fund, and other investors, the team is building an open-source, multithreaded proxy designed to make Postgres horizontally scalable without forcing developers to rewrite their application code.

How PgDog Handles the Heavy Lifting

The core philosophy behind PgDog is simple: keep the Postgres you know and love, but drop a smart proxy in front of it. Instead of managing complex sharding logic inside your application or wrestling with heavy database extensions, PgDog acts as an intermediary.

To get started, developers can pull the project's Docker image, point their application's DATABASE_URL to the proxy, and let PgDog handle the routing. Because it is written as highly concurrent, multithreaded code, it is designed to maximize CPU utilization without introducing heavy external dependencies or hidden serverless costs. This makes it highly portable; you can run it on a local laptop, in a colocation rack, or across a public cloud provider.

Born from Production Fire

PgDog isn't just a theoretical exercise in database design. The architecture is heavily informed by real-world scaling emergencies. The startup's co-founder, Lev, previously ran Postgres at Instacart during the massive demand surge of April 2020. As the company scaled 5x in a single month, the engineering team had to figure out how to serve hundreds of thousands of grocery delivery orders per minute.

The solution at the time involved manually sharding Postgres across Amazon RDS, Aurora, and EC2. PgDog is essentially the productization of those hard-won lessons, built from first principles to automate the sharding and scaling patterns that large engineering organizations usually have to build by hand.

Traction and the Road Ahead

Despite its small team size, PgDog is already seeing significant adoption. The project currently serves more than 2 million queries per second in production across dozens of deployments, managing over 20 terabytes of sharded data. On the open-source front, the project has racked up over 1.4 million Docker pulls on GitHub, supported by a weekly release cycle that ships new updates every Thursday.

With $5.5 million in runway, the startup is focusing on making the tool easier to operationalize for larger teams. While the core proxy remains open source, the team is actively developing an Enterprise edition of PgDog. This commercial offering is aimed at simplifying deployments on AWS and will come bundled with SLA-backed support directly from the creators.

For developers tired of the NoSQL pivot, PgDog offers a compelling promise: the reliability of Postgres, paired with the horizontal scale of a distributed database, all managed through a simple connection string swap.