RoachFest London Preview: Exploring Real-World Performance

There’s a certain kind of engineering story that only comes out under pressure — the real stories that reveal themselves when systems are stressed, customers are demanding, and the infrastructure either bends or breaks.

RoachFest 2025 is about honoring these stories, and providing space for enthusiasts and experts to gather and share what they’re building, challenges they encountered, and how they fixed it. Cockroach Labs has been hosting our annual database conference since 2022, and I’m honored to MC RoachFest London again this year – firstly because it means I didn’t muck it up too much last year. But also because of what the event represents, particularly in 2025.

This year’s conference will cover a new focus area which we are calling “Performance Under Adversity” — not the hypothetical kind that lives on whiteboards, but the real-world kind that defines the difference between being just available and being truly resilient. It’s the difference between being on-call and woken in the middle of the night because something went VERY wrong, and having the peace of mind to know that your system can recover by itself.

RELATED

Check out this video I recorded on “Chaos Testing CockroachDB.” I put CockroachDB through pod failures, network partitions, corrupted data, restricted bandwidth, time faults, and so much more:

As someone who spends most of his days either talking to engineers across industries — from fintech giants to gaming innovators — or finding solutions to tricky database problems, I’ve seen just how far the database world has come. But I’ve also seen how much further we need to go to build systems that can thrive in today’s fast-moving world. RoachFest is where we unpack that journey with the community.

To learn more about what to expect at RoachFest London 2025, check out our website. Register today, while seats last!

Why Database Benchmarking is Broken and What Comes Next

I’ve written before about the way distributed systems force us to rethink assumptions, especially when it comes to testing and architecture. And nowhere is that more obvious than in the world of database benchmarking. The traditional benchmarks we’ve relied on for decades are increasingly unfit for today’s businesses. They tend to overfit for ideal conditions: single-region deployments, uniform workloads, and tidy failure models.

But modern enterprises don’t live in ideal conditions. They’re global by default. Their workloads spike unpredictably. Their users don’t wait for maintenance windows or consult your business continuity plan before clicking “Buy Now.”

For example, let’s say you have a database that can process ~1M TPS, and you have another database that can process 2x more transactions per second, but costs 10x as much, well the first database offers a better price for performance. None of this matters, however, if the database goes down: that database will process 0 TPS.

To learn more about operational resilience, check out this video covering all things resilience, including DORA, RTO, RPO, and more:

Let’s also remember that the principle behind the CAP theorem was published back in 1999 and distributed SQL databases didn’t arrive until CockroachDB launched its commercial product in 2017. Fundamentally, the CAP theorem doesn’t do enough to describe the world in which distributed SQL exists. PACELC, first described in 2010, is more comprehensive and addresses these limitations, but our leadership felt that more needs to be done.

So Cockroach Labs is stewarding a new way to truly measure database performance.

At RoachFest London 2025, you’ll hear the full story from Peter Mattis, co-founder, CTO, and CPO of Cockroach Labs. Check out the full agenda on our website.

Real-World Lessons at Scale: From Payments and Gaming to Order Management and Regulations

A key part of RoachFest is customer stories. Admittedly, I’m most excited about the customer talks – not just because I used to be a CockroachDB customer – but because their learnings become my learnings. As a customer, I always felt that the Cockroach Labs team wanted to hear the good, the bad, and the ugly from me. That’s what the customer talks are all about. You’ll hear the actual reality that CockroachDB customers work with every day so you can make the most informed decision if you’re considering adopting distributed SQL or CockroachDB.

This year’s lineup features stories from some of the most forward-thinking engineering teams out there. Take SumUp, for instance—a company building payment solutions for millions of small businesses around the world. They’re using CockroachDB to power a platform that needs to be always available, always fast, and always secure. Their session will dig into how they architected for scale without sacrificing developer velocity.

We’ll also hear from Booking.com, who’ll share how they’re using CockroachDB to modernize and simplify legacy systems. It’s not just about shaving off milliseconds; it’s about removing complexity and making systems that are easier to understand, maintain, and evolve.

The team from Kaizen Gaming, which serves over 13 million customers across the globe, will also cover why they chose to run CockroachDB on OpenShift for their wallet application, which supports all of their revenue generating services. As the company continues to expand, they needed a database that could scale with them, while maintaining performance requirements.

And because no conversation about modern resilience is complete without addressing regulation, governance, and data sovereignty, we’ve got sessions dedicated to navigating them in 2025. The rules are changing fast, and resilience not only means technical uptime, but staying compliant, secure, and adaptable in a constantly shifting legal landscape.

Learn With and From the Experts

This year, RoachFest is bringing hands-on workshops to our London and Bengaluru events, courtesy of our in-house education team, the lovely people behind Cockroach University.

Besides MCing in London, I’m running the “Introduction to Distributed SQL” workshop, and I really do believe the best way to learn is by getting your hands dirty. We were the first distributed SQL database on the market, and everyone at the company feels a deep responsibility to teach the world about distributed SQL. If you’re already familiar with distributed SQL and want to dive into multi-region, I cannot recommend my colleague, Alistair Parry’s workshop on multi-region database architecture enough.

In all the webinars and events I run, I always like live questions – the harder, the better. At RoachFest, we’ll have lots of CockroachDB experts on-hand so it’s the very best place to ask questions. If we end up veering off course because someone’s asked something really interesting, then that’s a win too.

Why a Distributed SQL Conference Should Matter To Everyone

I’m not just excited about the speakers or the venue, I’m excited because this event is about more than database tech. It’s about shared learning and getting candid and technical with the people who are shaping the future of databases.

So, if you care about building systems that can handle anything the world throws at them—systems that scale, that heal, that stay up even when everything else is going down—(or if you just want to meet someone who was so passionate about a database that they got the logo tattooed) then you should be at RoachFest.

Come for the tech. Stay for the community. Hope to see you there!

Rob Reid is a technical evangelist at Cockroach Labs and the MC of RoachFest London 2025. Check out his author page for a full bio.