Cockroach Labs' Bold Vision for the Future of Cloud Databases

With the announcement of its serverless database, CEO Spencer Kimball looks ahead to what's next.

Cockroach Labs, named after an insect that has evolved over 300 million years, has shared its vision of the future evolution of databases—they will be “cloud agnostic” with less complexity, more resource elasticity, and global scale.

I attended Cockroach Lab’s RoachFest last week at the company’s New York headquarters, where I talked with CEO Spencer Kimball and CMO Peter Guagenti and heard from several customers.

Founded in 2015, Cockroach Labs now has nearly 500 employees. Last December, the company announced $278M in Series F funding, raising its valuation at the time to $5 billion. CockroachDB is a SQL relational database with a distributed architecture that gives the company its name—able to survive anything, like its namesake.

A handful of customers shared how they are using CockroachDB. JP Morgan Chase uses it to offer database-as-a-service internally. At Netflix, CockroachDB is deployed within its Device Management Platform (see the case study here). Other customer speakers included Shipt, the same-day delivery company owned by Target, and Devsisters, a Korean gaming company.

CockroachDB Serverless is now GA

On the product front, Cockroach made three announcements:

1. A serverless version of CockroachDB is now generally available. Last year, I reported on Cockroach’s plans to introduce a serverless database. The initial tier was a “forever free” version for pilot projects and test drives. Now, it’s GA with consumption-based pricing, which means customers can configure a CockroachDB database that automatically scales up and down, choose the region they want to deploy in, and set a monthly spend limit to avoid cost overruns. During setup, customers can choose to run the database on AWS or Google Cloud, with Microsoft Azure to be added as an option next.

2. Cockroach also introduced a data migration tool called MOLT (an acronym for Migrate Off Legacy Technology), which works with AWS’s Database Migration Service to load data into CockroachDB-on-AWS. The new tool automates schema conversion and data loading, which can be a slow and tedious process. Primary use cases are migrations from Oracle, PostgreSQL, and MySQL to CockroachDB. MOLT is in preview release.

3. For developers, there’s new integration with Vercel, a popular platform and tools for front-end Web apps, and HashiCorp Vault’s Dynamic Secrets feature for managing database credentials and encryption keys.

4 deployment options

From my perspective, CockroachDB Serverless was the star of the show because it’s foundational to Cockroach Labs’ go-forward strategy and, more importantly, it gives IT teams cloud database elasticity. That includes the ability to scale up in less than a second and scale down “to zero” for applications and workloads not in use.

So CockroachDB is now available in four deployment options:

• Dedicated - a fully-managed CockroachDB cluster, available now on AWS or Google Cloud, and eventually MS Azure.

• Serverless - fully managed, automatic scaling with free tier for starters.

• Self-hosted - for deployment in the customer’s data center or self-managed in the cloud.

• Core - an open source version of Cockroach DB without some of the enterprise functionality such as geo-partitioning and geo-located backup.

This webpage shows how the four versions compare. Every cloud database provider should have a page like this.

What’s next?

That’s what’s new; here’s what’s next.

During his keynote presentation, Cockroach CEO Spencer Kimball pointed to a few trends that he expects to lead to “a more unified market” over the next few years. They include: reduced complexity as more capabilities are delivered as a service; the serverless model with its higher level of resource abstraction (i.e. compute & storage); consumption pricing; and geographically distributed systems and data.

Beyond serverless, Kimball said, is infrastructureless—with the same principles of elasticity throughout the tech stack for things like load balancing across availability zones. Next-generation infrastructure for next-generation applications.

The other big thing is multi-cloud. Kimball used the terms heterogeneity and “cloud agnostic,” which may be a different ways of saying multi-cloud, although it also hints at something more than just making a database available in two or three clouds.

In other words, it’s when you think about how these trends come together in the years ahead that things get really interesting—serverless databases with consumption pricing, with built-in resilience and load-balancing, across clouds and regions. For Cockroach Labs, the ultimate vision is what Kimball referred to as a “CockroachDB global service.”

Multi-region, multi-cloud

Cockroach Labs still has some development to do with its platform to get to that future state. Next on its roadmap is multi-region serverless, to enable this same kind of database elasticity across regions.

Also, Kimball told me that Cockroach sees interest among customers is deploying CockroachDB serverless within their own data centers as a way of offering scale up/scale down database-as-a-service to their own dev teams and lines of business.

I’ll wrap up with a few additional points on multi-cloud. In a recent LinkedIn post, I noted that the industry, in the wake of Oracle’s emerging multi-cloud strategy (see post below), will be talking a lot more about multi-cloud in the days ahead.

Larry Ellison Gets Serious About Multi-Cloud. What Is Oracle's Next Move?

Cockroach Labs is already well along the path to multi-cloud, with Cockroach DB available on AWS and Google Cloud, and Microsoft Azure due next. When I asked Kimball how he sees this unfolding, he said that multi-cloud is not only something that customers want and expect as an option, but increasingly a requirement in regulated industries.

If you think of that as the direction the industry is heading, Kimball said, it’s only a matter of time before multi-cloud deployment becomes even more seamless and automated. For example, he said, you could potentially click a button and move from AWS to Google Cloud. Or deploy a multi-cloud cluster across all three major clouds for maximum replication and resiliency.

---

To keep up with all of my observations and join me in the conversation around cloud databases, follow me on LinkedIn.