February 8, 2025, holds special meaning for Cockroach Labs – it marks our ten-year anniversary. Reflecting on this milestone, I’m filled with gratitude and a sense that we are helping to shape the future of databases. A decade of continuous innovation – and exponential growth in data-under-management – is no small feat, and it wouldn’t have been possible without the trust of our pioneering customers and the dedication of our exceptional team around the globe.
Databases are the lifeblood of every modern business, ceaselessly laboring behind the scenes to power the world’s applications and services. It has been an intense labor of love building a paradigm-shifting database that exploits the incredible benefits of a distributed architecture (e.g., resilience, scale, geo-locality), but also supplies the myriad features necessary to migrate legacy enterprise data architectures. A decade of focus, iteration, and relentless problem-solving has made CockroachDB a solution that bridges from venerable but outdated technologies to a future where applications are built above – and even actively across – private and public clouds. CockroachDB delivers always-on customer experiences that can accommodate success at any scale, across the globe.
Celebrating Cockroach Labs' 1st birthday with an ode to our namesake.
From Day One, our goal has been to bring together the benefits of both SQL and NoSQL, using a cloud-native, distributed architecture. But what we have accomplished for our customers over the past 10 years is only the beginning. We’re most excited now for the next decade of innovation, built on the immensely powerful foundation that our distributed architecture provides. While legacy databases struggle to match CockroachDB’s current capabilities, we aim to continuously bring game-changing new technology to our customers. This is the essence of the hard-won success that has shaped Cockroach Labs – enabling our customers’ success is the ultimate prerequisite for our success.
CockroachDB gives you another option that’s not strictly SQL and not strictly NoSQL. It’s [a] next-generation database that helps us solve a lot of challenges as we become more dependent on data and need to scale out further and further.
- Bryan Call, Senior Principal Engineer, Route
3,650 Days of Distributed SQL and Counting
At Cockroach Labs, our mission has always been clear: simplify how businesses build and operate world-changing applications.
From the start, our guiding principle has been to “Make data easy,” a vision that has shaped CockroachDB into a database that solves critical challenges in data distribution, scalability, and resilience. Global brands like Bose, Hard Rock Digital, Shipt, Checkout.com, and Fortinet rely on CockroachDB to power some of their most important applications across cloud, on-premises, and hybrid environments.
The reason we selected [CockroachDB] in the first place was that it was an obvious choice to be able to cut away a whole lot of engineering effort and time, and today we still see that has paid off.
- Matt Stephenson, Senior Principal Software Engineer, Starburst
One of the most pivotal decisions in our early days was to build a relational database (that speaks SQL) instead of opting for a simpler NoSQL approach. This was the vision of our Co-Founder and Chief Technology and Product Officer, Peter Mattis, who saw clearly the benefit of going straight at the hardest problem in the industry, and setting ourselves apart in a competitive market. SQL has long been the choice for mission-critical, complex workloads, and making it work well – especially with a distributed architecture – dramatically upped the level of difficulty of what we set out to achieve ten years ago. While challenging, that decision allowed us to support workloads for some of the world’s most demanding businesses.
At our most recent customer conference, RoachFest 2024, Peter and I spoke about the need for “Redefining Resilience in Today’s Modern World,” a presentation that highlighted some of the innovation that has allowed CockroachDB customers to do remarkable things, impractical or even impossible in the annals of database history.
CockroachDB customers routinely upgrade versions, “repave” nodes, scale clusters up and down, and alter schemas – under peak load. Customers seamlessly migrate from one cloud vendor to another with zero application downtime. Customers even actively replicate across public cloud vendors, allowing a cloud vendor to be “turned off” completely, while maintaining complete business continuity! It’s the regular success stories like these that make the incredible efforts over the past decade worth every sleepless night.
What’s next for Cockroach Labs and distributed SQL?
What’s unique about building a relational database is that reaching ten years is not a time to rest on any laurels; it’s a time to accelerate. While many businesses would be looking at building additional products to complement their initial offering, Cockroach Labs must continue to relentlessly tackle the next set of difficult problems facing databases. Every application in the world requires an operational database, and the state of the art never stops advancing.
We’re looking forward to an exciting year ahead improving CockroachDB’s pgvector support so that operationalized AI use cases can be built on the shoulders of CockroachDB’s powerful differentiators. After all, it’s straightforward to see how the intelligence increasingly being built into applications must be just as resilient, scalable, and geo-aware as the other metadata served from the database. Making vector search scale to many billions of vectors, maintain accuracy and freshness – and brook no compromise on consistency with other metadata – is a challenging task at the forefront of database research.
We introduced pgvector support in Fall 2024 to power AI use cases.
Our most recent release, CockroachDB v25.1, continues our focus on enhancing resilience and performance at scale, while driving down the cost to serve. In this release, we’ve achieved a nearly 20% performance improvement, making it more cost efficient than ever. In addition, we’ve implemented Leader leases (in Preview), which speed up recovery from network partition failures and liveness failures. These features are directly driven by the needs of our customers, whose goals we are proud to support through CockroachDB.
Perhaps even more important to CockroachDB customers than new functionality (and even more difficult to get right) is increasingly bulletproof resilience – low-variance performance despite unpredictable patterns of usage and even concurrent cluster maintenance operations. Benchmarks are well and good, but no database runs in the real world under ideal conditions. What CockroachDB aims to perfect is its own behavior under the most challenging conditions – scenarios which would be classified as “disaster recovery conditions” for virtually every other database on the market. Not only does CockroachDB increasingly shift the terrain for those scenarios from “disasters” to “IT resilience”, but it seeks to maintain nominal performance and latency profiles. Our work likely will never be done on this count, but we expect to keep pushing the envelope and redefining the industry standard for mission-critical applications.
CockroachDB already has “global tables,” which provide low-latency local reads across globally-distributed clusters, while maintaining consistency. I’m only being partly facetious when I say I’m looking forward in the next 10 years to seeing the first inter-planetary database cluster, guaranteeing consistency between the Moon, Earth and Mars!
To our customers, employees, and partners: Thank you for believing in Cockroach Labs and for joining us on this exhilarating journey. Here’s to the next chapter—and to building a foundation that stands the test of time.
