Postgres start-up ecosystem 2024

Posted on So 10 März 2024 in Blog

Postgres start-up ecosystem 2024

This is the English version of an article published in February. For the German version go to Postgres Start-up Ökosystem 2024.

A conversation with a colleague about databases and the importance of the longevity of data gave me the idea, to investigate the postgres start-up ecosystem. I talked with him about the NoSQL hype and how few databases from this era really still exist today. So the right choice of a database is important. In my opinion, there has been something of a gold-rush atmosphere in the Postgres ecosystem in recent years and there are therefore numerous start-ups in this area. Not all of them will probably manage to live forever and so I thought I would take an (incomplete) snapshot of the current status. Therefore, some companies are listed in the following section. However, the list does not claim to be exhaustive.

Update: 2024-04-01: Xata was added.

Inhalt:

Timescale

One start-up that has already been a guest on this blog with its extension is Timescale (formerly TimescaleDB). In the German blog post Podman Container Logs mit Fluent Bit, Postgres und TimescaleDB verarbeiten, I already described some time ago how I use Timescale. It is an extension that specializes in the storage of time series data. Timescale offers some very nice features out-of-the-box, such as a data retention policy, compression and continuous aggregates. In my German Wochennotiz Nr. 8 I already referred to the deepdive of a German podcast called programmier.bar, but here it fits again: TimescaleDB mit Sven Klemm.

In the last funding round (Series C) Timescale raised 110 million dollars: Year of the Tiger: $110 million to build the future of data for developers worldwide. In a Series B round they already raised 40 million dollars (Timescale grabs $40M Series B as it goes all in on cloud version of time series database).

Supabase

Supabase describes itself as an open source Firebase alternative and is a backend for web and mobile applications. In addition to real-time communication, Supabase also offers edge functionality. APIs are provided via PostgREST or GraphQL.

Supabase raised 80 million dollars in the last funding round (Series B) (Supabase Series B) and has already raised 30 million dollars before that (Series A) (Supabase $30m Series A).

Supabase was also part of the Y Combinator Program Summer 2020: Supabase - Build in a weekend. Scale to billions.

Neon

Neon is a serverless Postgres variant. Since its incorporation, some well-known personalities from the Postgres environment such as Heikki Linnakangas, Stas Kelvich and Anastasia Lubennikova have been involved. Peter Geoghegan, on the other hand, left Neon after a relatively short time and is now at AWS.

Probably the best feature of the serverless architecture is the fast spin up of a new Postgres instance. Neon also offers branching, as we know it from git.

Neon received 46 million dollars in the last funding round (Series B) and 30 million dollars in a Series A round: We Raised another $46M – What’s Next?.

Tembo

The goals of Tembo are to end database sprawl and create a developer-friendly platform. Tembo has set itself the goal of making it easier to benefit from the ecosystem of Postgres extensions and offers many of these extensions on its own platform. The main product is called Stacks and is a use case specific auto-tuning Postgres instance.

I first came across Tembo at last year's pgconf.eu in Prague, where they presented their own message queue extension: Blazingly Fast Message Queue on Postgres with Rust. With David E. Wheeler they also have a quite well known person from the Postgres environment on board.

Tembo has raised 6.5 million dollars in a first round (Introducing Tembo).

FerretDB

FerretDB calls itself a truly open source MongoDB alternative, built on Postgres. They attempt to create a MongoDB compatible product that accesses Postgres under the hood. Since February 14th there is the possibility to use FerretDB via Tembo Cloud (MongoDB capabilities on Postgres with Managed FerretDB on Tembo Cloud).

A few years ago, there has already been an attempt by another company to develop a MongoDB compatible database. The name was ToroDB and at least the github repository still exists: https://github.com/torodb/server.

FerretDB has completed a seed round with 2.8 million dollars.

Lantern

Lantern has developed an alternative to pgvector and released its own extension to use Postgres as a vector database: Lantern.

Lantern is in the Y Combinator Winter 2024 program: Lantern - PostgreSQL vector database extension to build AI applications. For taking part of the program, there was a funding of 500,000 dollars from Y Combinator.

Ubicloud

Ubiclud is a relatively young company that I first came across at pgconf.eu in Prague. At pgconf.eu I was told that some of the founders also founded the Company behind the Citus sharding extension. Ubicloud describes themselves as an open, free and portable cloud that, at least in Prague at the time, primarily ran on Hetzner servers. According to the website, OVH and AWS Bare Metal also appear to have been added as alternatives.

Ubicloud is in the Y Combinator Winter 2024 program: Ubicloud - Open source alternative to AWS. In the course of this, there was a funding of 500,000 dollars from Y Combinator.

PeerDB

PeerDB enables data from Postgres to be streamed quickly and efficiently into data warehouse solutions like Snowflake. To track changes in Postgres, the Change Data Capture (CDC) mechanism of Postgres is used, among other things. I already mentioned PeerDB briefly in my German Wochennotiz 11.

PeerDB was in the Y Combinator program summer 2023: PeerDB - Fast, native ETL for Postgres.

ParadeDB

ParadeDB describes itself as an Elastic alternative built on Postgres. I already mentioned ParadeDB briefly in my German Wochennotiz 9. I mentioned the possibility of using Postgres as a fast OLAP database with ParadeDB.

ParadeDB was in the Y Combinator program summer 2023: ParadeDB - Postgres for Search and Analytics.

ReadySet

ReadySet is a realtime caching layer for Postgres and MySQL. I already mentioned ReadySet briefly in my German Wochennotiz 11. ReadySet works with IVM (Incremental View Maintenance) and Logical Replication to keep changes to the data in the cache.

ReadySet has raised 29 million dollars in a Series A funding round (ReadySet Raises $29M in Series A Funding).

OrioleDB

OrioleDB utilizes the extensibility of Postgres and tries to fight bloat with a new storage engine. The current implementation of the Multi-Version Concurrency Control (MVCC) is susceptible to table bloat. OrioleDB is also backed by Alexander Korotkov, a well-known name in the Postgres world.

Supabase has invested 500,000 dollars in OrioleDB.

There has already been an attempt to solve the problem with a project called zheap.

ZomboDB

ZomboDB creates a connection between Postgres and Elastic by using Elastic as a new index type in Postgres.

The information on ZomboDB itself is somewhat meager, so the description is brief.

EdgeDB

EdgeDB relies on the Postgres query engine and data storage, but uses its own query language. EdgeDB calls itself a graph-relational database.

EdgeDB has received a total of 19 million dollars in funding: EdgeDB raises $15M ahead of the launch of its cloud database service.

Nile

Nile is working on a serverless version of Postgres. The founders were previously at Confluent, the company behind Kafka. When asked why they rely on Postgres:

Why Postgres? According to Subramanian, that’s where the market is going. Virtually every company the team talked to was betting on Postgres. “It’s pretty obvious that it’s going to become the Linux of databases,” he said. The fact that it’s open source and easily extensible also enables Nile to do what it does.

Nile raised $11.6M in a seed round: Nile raises $11.6M in seed funding to build a Postgres-powered data layer for SaaS applications.

Hydra

Hydra is working on improving the analytical capabilities of Postgres and is repeatedly mentioned as an open source alternative to Snowflake (Hydra aims to be the open cloud data warehouse alternative to Snowflake).

They offer an extension, Hydra Columnar, and have developed their own extension manager: pgxman. In addition, they also offer an extension for working with DuckDB: pg_quack.

Hydra was in the Y Combinator program winter 2022: Hydra - Query billions of rows instantly on Postgres..

PostgresML

PostgresML describes itself as a complete MLOps platform in a PostgreSQL extension. Two extensions are (co-)developed: PostgresML and PgCat (Connection Pooler, like pgBouncer).

Recently they raised some money: PostgresML raises $4.7M to launch serverless AI application databases based on Postgres.

pgEdge

pgEdge describes itself as Fully distributed PostgreSQL, optimized for the network edge. According to its own descriptions, pgEdge is also a true multi-primary distributed database system. They developed an extension to offer multi-primary replication: Spock.

The company was founded by two veterans from the Postgres environment, who also worked for EDB, for example.

pgEdge has completed a seed round with 9 million dollars: Distributed database provider pgEdge emerges from stealth with $9M.

postgres.ai

postgres.ai enables fast cloning of Postgres databases to accelerate CI/CD processes.

The company was founded by Nikolay Samokhvalov, who runs the Postgres podcast Postgres.FM together with Michael Christofides.

Xata

According to a founder profile series (SV Angel Founder Profile Series: Monica Sarbu — Xata) Xata is described as following:

Xata offers a serverless database service that combines the power of a traditional database with the UX of a spreadsheet.

Xata had a $30 millions Series A funding round (Database Startup Xata Steps Forward With An All-Women Board And $30 Million In Fresh Funding) and focuses on being a serverless database service for Jamstack apps.

Former Start-ups

As this is the first edition of the start-up ecosystem, it is not yet possible to see how the start-ups have changed since the last edition, so I have mentioned a few companies that came to mind quickly.

One of these companies is Swarm64, a company acquired by ServiceNow. The focus there was on the use of Postgres as a data warehouse. I immediately remembered at least one episode of a German podcast where Swarm64 was reported on: Techtiefen #33 Postgres als DWH.

Another start-up that was acquired is Citus. Citus has developed a sharding extension for Postgres, which is probably already familiar to some. Citus was acquired by Microsoft a few years ago. The extension is still available as an open source. And the website still exists: citusdata.com.

Postgres