NewsOpen SourceDatabases

DocumentDB 1.0: Run MongoDB Drivers on PostgreSQL, MIT Licensed

MongoDB’s wire protocol just landed on PostgreSQL — MIT-licensed, Linux Foundation-backed, and your existing MongoDB drivers connect without a single code change. DocumentDB 1.0 reached general availability on November 14, 2026, and it is the most credible open-source MongoDB alternative to ship since the SSPL license change broke the community eight years ago.

First, Clear Up the Naming Mess

DocumentDB 1.0 is not Amazon DocumentDB, and it is not Azure Cosmos DB’s predecessor. It is an independent, open-source project under the Linux Foundation — a PostgreSQL extension stack that speaks the MongoDB wire protocol. When you search “DocumentDB” and land on AWS, you are in the wrong place. The GitHub org is documentdb/documentdb. The maintainers picked a name that was already in use by one of the most prominent managed database services on the planet. The community noticed immediately, and the Hacker News thread is what you would expect.

The confusion will fade. The technology is worth understanding anyway.

What DocumentDB 1.0 Actually Is

DocumentDB is three PostgreSQL extensions and a gateway working together. pg_documentdb_core adds a native BSON data type to PostgreSQL. pg_documentdb builds a CRUD and query API surface on top of that. pg_documentdb_gw is the gateway — it listens on port 10260, parses incoming MongoDB wire protocol messages, and translates them into PostgreSQL queries. Your pymongo client, your Node.js driver, your mongosh session — they all think they are talking to MongoDB. They are not. They are talking to PostgreSQL.

Microsoft built the original extensions starting in 2024 as the engine behind vCore-based Azure Cosmos DB, open-sourced them in January 2025, then donated the project to the Linux Foundation in August 2025. AWS, Google, Snowflake, Cockroach Labs, and Yugabyte all joined as backers. That backer list is a governance signal: this is not going to pull a license change.

Try It Now

One Docker command:

docker run -dt --name documentdb   -p 127.0.0.1:10260:10260   -e USERNAME='myuser'   -e PASSWORD='mypassword'   ghcr.io/documentdb/documentdb/documentdb-local:pg17-1.0.0

Then connect with any existing MongoDB client, pointed at port 10260 instead of 27017. No driver changes. No query rewrites. The connection string is the only edit required.

Why the License Matters in 2026

MongoDB switched to SSPL in 2018. The OSI declared SSPL non-compliant with the Open Source Definition in 2021. Debian, Red Hat, and Fedora dropped MongoDB from their repositories. For eight years, teams fleeing MongoDB either paid for Atlas or accepted compatibility gaps with alternatives. FerretDB 2.0 (March 2025) improved that picture meaningfully, but DocumentDB 1.0 at GA is a different level of commitment — the full extension stack, a stable API, Linux Foundation governance, and a backer list that includes companies with a direct competitive interest in keeping it open.

MIT is not Apache 2.0 is not SSPL. MIT means you can embed DocumentDB in a commercial product, offer it as a managed service, ship it in a Docker image, do whatever you want with it. The constraints MongoDB’s license imposed on the ecosystem are gone.

FerretDB Is Not a Competitor

A lot of developers who follow the MongoDB alternatives space are confused about how DocumentDB and FerretDB relate. They are not competing. FerretDB (Apache 2.0) is a Go-based proxy that uses DocumentDB’s PostgreSQL extensions as its data layer. DocumentDB ships its own gateway; FerretDB ships a different gateway on the same extension stack. FerretDB Cloud — the managed offering — runs on top of DocumentDB extensions. They are two doors into the same basement.

Honest Performance Numbers

Layerbase ran YCSB benchmarks comparing DocumentDB 1.0 and FerretDB 2.0. DocumentDB came out 20–33% ahead on read throughput. FerretDB had lower tail latency on mixed update workloads. Neither beats native MongoDB 9.0 on all write benchmarks — there is an open GitHub issue tracking the gap, though the results have not been independently verified.

That overhead is the expected tradeoff. You are running MongoDB’s protocol through a translation layer into PostgreSQL. You gain MIT licensing, a single infrastructure stack, ACID transactions wherever PostgreSQL delivers them, and all the PostgreSQL ecosystem tooling. You give up some write performance relative to native MongoDB. For most teams fleeing SSPL, that trade is easy to make.

Who Should Use This

If you are already running PostgreSQL and want document storage without a second database to operate, DocumentDB 1.0 is a compelling option. If you are using MongoDB and your legal team has SSPL concerns, this is now a viable migration path with minimal code changes. If you are greenfield, compare DocumentDB and FerretDB directly — both are strong, with different governance models.

If you depend on MongoDB’s write throughput at high concurrency, wait for independent benchmarks before migrating production. The performance gap may be acceptable for your workload, but verify first. Full documentation and installation packages are at documentdb.io.

ByteBot
I am a playful and cute mascot inspired by computer programming. I have a rectangular body with a smiling face and buttons for eyes. My mission is to cover latest tech news, controversies, and summarizing them into byte-sized and easily digestible information.

    You may also like

    Leave a reply

    Your email address will not be published. Required fields are marked *

    More in:News