Open-source databases in the enterprise: freedom without the 3 AM pages

Open source isn't free. It's prepaid engineering.

I say that as someone who has built a business on open-source databases and believes in them completely. PostgreSQL, MySQL, Apache Cassandra, MongoDB, Redis, Kafka, Elasticsearch: this is world-class software, running the largest systems on the planet, with no license invoice attached.

This is not niche any more.

PostgreSQL is now the most-used database among developers, two years running, and it led all engines for growth in H1 2026; Cassandra powers write-heavy systems at Netflix, Uber, and Discord. The enterprise question is no longer whether to run open source, but how.

But every enterprise that adopted open source purely to cut costs has learned the same lesson, usually some time in the night: the license fee didn't disappear. It converted into engineering hours. Somebody still has to run the thing.

The companies that win with open source are the ones who understand what they're actually buying. So let's be precise about it.

What open source actually buys you

Not "free software." Three more valuable things:

Freedom to leave. 

No vendor controls your runtime. If your provider disappoints you, you change providers, not databases. The skills are everywhere, the tooling is portable, and your data sits in formats a thousand companies can operate. That leverage quietly improves every negotiation you'll ever have.

Freedom to see. 

When something behaves strangely at the storage engine level, the answer isn't a support ticket into a black box. The code is right there. For business-critical systems, the difference between "we can find out" and "we've asked the vendor" is measured in hours of downtime.

The world's largest engineering team. 

PostgreSQL alone gets more expert attention in a year than any proprietary engine's vendor could staff in a decade. Bugs get found, fixes ship, extensions bloom. You're not betting on a company's roadmap; you're drafting behind an ecosystem.

What open source does not buy you

Nobody to blame, and nobody on shift.

When the commercial database breaks, you call the vendor; it's their problem, contractually. When PostgreSQL breaks, it is entirely, completely your problem. The community will help you, but they won’t fix it.

So the honest total cost of open source is: zero license fees, plus the fully loaded cost of the people who provision it, tune it, patch it, back it up, test the restores, watch it overnight, and answer if something goes wrong. For a serious production estate, that's not a part-time job bolted onto a developer's week. Cassandra makes the point: it ships with no built-in repair scheduling and punishes mistuned compaction with silent, creeping degradation, so an unstaffed cluster doesn't fail loudly, it rots quietly. Pretending the operations are free: that's where the 3 AM stories come from, and why some teams retreat to expensive platforms and conclude "open source didn't work for us."

The software worked fine. The staffing model didn't.

The third option nobody mentions in the build-vs-buy meeting

The debate is always framed as two doors.

Door one: proprietary platform, pay the license and the markup, get the operations included, hand over control.

Door two: open source, keep the freedom and the savings, and staff a 24/7 operations capability yourself.

There's a third door: open-source engines, running in your own cloud accounts, operated by someone else's reliability engineers.

You keep everything open source promised: no license fees, no lock-in, full visibility, infrastructure at your own negotiated rates. And the 3 AM problem belongs to a team that's awake anyway, because that's their whole job.

That third door is what Maxima Consulting builds. Managed open-source databases, from PostgreSQL and MySQL to Apache Cassandra and 40+ engines besides, in client accounts, operated 24/7/365 by follow-the-sun SRE teams from Kraków and Pune under a 99.99% uptime SLA enforced end to end, at fixed operational pricing.

Our Cloud Orbit platform, built on industry-standard open-source tooling, is the machinery underneath.

The incentive alignment is the part I like most: because the stack is open and lives in your accounts, you can fire us without a migration. The only lock-in available to us is being good.

When proprietary is still the right call

Balance, because this audience deserves it: if a proprietary engine has a capability your workload genuinely needs (some do), or your organization has decades of tooling and skills wrapped around one, ripping it out to make an ideological point is bad engineering. Open source is a default, not a religion. Make the exceptions consciously, and price the exit before you enter.

FAQ

Are open-source databases reliable enough for enterprises? 

Yes; PostgreSQL, MongoDB, MySQL, Kafka, and Redis run many of the world's largest production systems. Reliability outcomes depend on the operating capability around the engine (monitoring, tested restores, capacity planning, 24/7 coverage), not on the license model.

What is the real cost of open-source databases? 

Zero license fees plus the full cost of operations: provisioning, tuning, patching, backups with tested restores, and around-the-clock incident coverage. The software is free; the reliability is engineered.

Should we self-manage open-source databases or use a managed service? 

Self-manage if you genuinely have (and want to keep funding) a 24/7 database operations capability. Use a managed service that operates in your own cloud accounts if you want open source's freedom and cost profile without staffing the pager.

Do managed open-source database services create new lock-in? 

Not structurally, if the engines are unmodified open source and the deployment is in your accounts: switching then means changing operators, not migrating data. Verify both properties before signing; they're the whole point.