Asset 20 8 2
Does AI recommend your business? Run the free check →

Join 15,000 business owners, marketers and entrepreneurs. The Sunday newsletter you'll be annoyed only arrives once a week.

Article

CockroachDB Alternatives (2026): 10 Databases and How to Choose

Switching databases twice is the outcome to avoid, because each migration means rewritten queries, retesting and a tense cutover. If you're leaving CockroachDB, first decide whether you need distributed SQL, plain Postgres or MySQL compatibility, or a managed cloud service, then prove one real workload on two candidates before you commit.

The right CockroachDB alternative depends on why you're leaving: if it's the licence, you want another distributed SQL database; if it's the weight of running one, a single Postgres or MySQL-compatible database will probably serve you better. Settle that first, because the two answers lead to different shortlists and very different migrations.

This guide is for engineers and engineering leads weighing a move. It argues by type of database rather than by brand, because the type you pick decides most of the work that follows, and it starts with a check that might save you the migration altogether.

Read the licence terms before you read anything else

Cockroach Labs set out the change in its own post, 'Evolving our self-hosted offering and license model', published on 15 August 2024. Self-hosted CockroachDB used to come in two flavours, a free Core version and a paid Enterprise version. From version 24.3 in November 2024 Core was retired, and the same structure was applied to new patch releases of 23.1 and later.

What replaced it matters for your decision. The post describes an Enterprise Free licence for individual developers, students, academic researchers and businesses under $10M in annual revenue, renewed each year, plus a 30-day Enterprise Trial. So if your company sits under that revenue line and the architecture suits you, check whether the free licence covers you before you plan anything. If you're over it, or you don't want a yearly renewal hanging over a core system, you've got a real reason to look around.

Pick the type first

Licence is the only problem and you need horizontal scale across regions: shortlist the distributed SQL databases with Postgres compatibility. Your mental model, failover expectations and much of your runbook carry over.

You never needed a cluster: go back to plain Postgres, self-run or as a managed service. Look at your last year of incidents. If none of them would have been prevented by surviving a regional outage, this is probably you, and the smaller operations load will feel like a relief.

You want nobody on your team patching databases: pick a managed service from the cloud you're already on, and accept the lock-in knowingly.

You're building an app and want the backend extras too: a Postgres platform with auth and APIs around it saves weeks of glue code.

You're on MySQL at heart and outgrowing one server: a sharding layer is the route, and my list of PlanetScale alternatives covers that path in more depth. If a document database is on the table instead, see these MongoDB alternatives.

The odd ones out on the list below, one built on SQLite for huge numbers of small databases, one for real-time analytics, one multi-model, aren't drop-in swaps. Test them only if your workload already looks like what they're built for.

Compatibility decides the size of the job

CockroachDB's own PostgreSQL compatibility page says it supports the PostgreSQL wire protocol and the majority of PostgreSQL syntax. That makes a Postgres-compatible target the shortest hop. A MySQL-compatible target means rewriting more than you'd expect, even if your ORM claims to hide the difference.

Plan for the work to sit in queries, drivers, your ORM and your schema migrations rather than in copying rows. And treat 'Postgres-compatible' on any label as a starting point. Before you get attached, list the SQL your app leans on hardest, especially transactions with retries, extensions and anything that reads system tables, and check each item against the target's own documentation.

The ten databases

Each one gets a line on what kind of database it is. They run by name, nothing more.

Amazon Aurora

A managed relational service from a big cloud provider, compatible with both MySQL and PostgreSQL, for teams who want someone else running the servers.

CrateDB

An open-source database built for real-time analytics on live operational data, worth a look when dashboards and queries over fresh data are the real workload.

Google Cloud Spanner

A fully managed database that combines relational, graph, key value and search in one service, with distribution handled for you.

PostgreSQL

The open-source relational database that much of distributed SQL builds on or copies; the natural landing spot if you never needed a cluster.

Supabase

A Postgres platform that wraps the database in auth, instant APIs, realtime, storage and vectors, suited to product teams shipping apps quickly.

SurrealDB

A multi-model database that scales to zero when idle and speaks both the PostgreSQL and MySQL protocols, for workloads that come in bursts.

TiDB

Distributed SQL you can run as a cloud service or self-manage, the closest in spirit to what you're leaving if horizontal scale stays essential.

Turso

A SQLite-based database built to give every user, agent or tenant a database of its own, a very different answer to multi-tenant scaling.

Vitess

A MySQL-compatible system with sharding built in, so applications grow past one server without carrying sharding logic in their own code.

Work with me

Want AI doing the heavy lifting in your marketing?

I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.

YugabyteDB

A distributed, Postgres-compatible database that can start serverless and grow to critical workloads, for teams keeping the distributed model.

What each type costs to run

No one publishes a figure you can compare across this list, because the units differ: some bill per vCPU, some per request or row read, some per hour of a node, and the open-source projects bill nothing at all. Here's the shape, from the vendors' own pricing pages checked in November 2026.

Free to start is common. Five of the pricing pages checked offer a free tier at $0 a month, with limits on storage, requests or active projects, and one of those pauses free projects after a week of inactivity. Where a flat first paid plan is listed, it ran from about $5 to $25 a month.

Dedicated distributed clusters are a different league. One managed distributed service lists $125 to $167 per vCPU per month; another starts its provisioned and dedicated tiers at between $400 and $1,800 a month for compute, with storage extra. For comparison, Cockroach Labs' own page puts a 2 vCPU, 100 GB cluster at about $203 a month and a 12 vCPU example at $1,528, and gives new cloud organisations a 30-day trial with $400 of credit.

Watch the transfer lines if you go multi-region. One distributed vendor's price list charges $0.01 per GB within a region and $0.02 to $0.08 per GB across regions, so a chatty multi-region setup costs more than the compute line suggests. Self-run open source has no licence bill, but the people and machines to run it are the real cost, and for a small team that's often the bigger number.

The practical step: put your current monthly usage through each shortlisted provider's own calculator, and add an honest estimate of engineering hours for anything you'd run yourself.

Managed, self-run, or both

The big cloud providers offer managed relational and distributed databases that take patching, failover and backups off your hands. Open-source projects let you run everything yourself, wherever your data has to live. Several options sit in between, with a managed cloud service and a self-managed version of the same engine.

If you self-host because of data residency or a regulator, keep that at the top of your requirements and cut anything that can't meet it. If you self-host only because there wasn't a managed option you trusted when you started, test one now. Managed services from a single cloud tie you to that cloud's regions and pricing, which is fine until you want to move again.

An hour-long failure drill

The question isn't whether a candidate runs your queries. It's what your app sees when a node dies mid-write. You can get a first answer in an hour.

Take two candidates of the type you've chosen and stand up the smallest cluster or instance each allows. Load a realistic slice of one real service's data, not a toy schema: the same indexes, the same row widths and the same awkward JSON column you've got in production.

Spend the first twenty minutes running your three heaviest queries and your busiest write path from your real application code, through your real ORM. Note anything that errors, needs rewriting or is mysteriously slower.

Spend the next twenty breaking things. Kill a node, or force a failover on the managed service. Drop the client's network for thirty seconds. Run a schema migration while writes are flowing. Watch what your app sees: errors, retries, stale reads, or nothing at all.

Use the last twenty to restore from a backup into a fresh instance and time it. That number matters more than any benchmark, because it's how long you'll be down on your worst day. You won't have a final answer after an hour, but you'll know which candidate earns a few weeks on a non-critical service.

Mistakes that make teams migrate twice

Choosing on a vendor benchmark. Your own failure drill beats any published chart.

Trusting compatibility on the label. Run the real application against the target, including migrations and background jobs, before you set a cutover date.

Cutting over in one go. Keep the old cluster live and in sync until the new one has carried real traffic through a full release cycle, including a deploy and an incident.

Losing the jobs the old database did for you. Automatic failover, multi-region writes, online schema changes and backup schedules all need a named owner in the new setup. Write them down and tick them off as each one is replaced.

Overcorrecting. A team burnt by one licence change can leap to the most exotic option on the list. Unless your workload matches its shape, the boring choice is the safe one.

Do you make a distributed or cloud database? Tool makers can ask to be considered for a featured place on this list.

Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.

Published and maintained by the Lilach Bullock team, covering marketing, AI and business growth.
Your buyers are asking AI who to use. Does it say you?

See for free whether ChatGPT, Claude, Perplexity, Gemini and Google name you, and get the plan to become the answer.

Check my AI visibility →
Sundays only

Get the Sunday newsletter.

One email a week. AI experiments, marketing tactics, and the workflows Lilach is building right now in her own business.

Subscribe free

Let’s get your marketing running on AI.

Book a free 30-minute call

We figure out what you need, where AI fits in, and what working together would look like.

Book the call →

Or take the 30-second calculator

You’ll see the hours and the money quietly leaking out of your week, and the three workflows worth building first.

Take the calculator →

Or grab the free AI resource library

Prompt packs, templates, checklists, and swipe files. The exact tools I build for paying clients. Yours, free.

Get the library →
Keep reading

More from the blog.