Academic Paper: How CockroachDB Cloud Delivers Affordable, Resilient, and Globally Scalable Databases

As applications scale globally, efficiency and cost control become top priorities. Elastic database tiers let data architects match resources to workload needs, ensuring consistent performance, predictable costs, and simplified operations at any scale.

CockroachDB Cloud offers three service tiers, Basic, Standard, and Advanced, that let users choose the right balance of flexibility, control, and cost. Basic and Standard are our multi-tenant offerings, designed for ease of use and affordability, while Advanced provides dedicated resources for larger-scale enterprise deployments.

So how can CockroachDB offer Basic and Standard, including a “free forever” option for small clusters, at a lower price point than Advanced, while still delivering the resiliency, scalability, consistency, and multi-region capabilities that CockroachDB is known for?

The answer lies in the multi-tenant architecture that powers these tiers, which we describe in our research paper, “CockroachDB Serverless: Sub-second Scaling from Zero with Multi-region Cluster Virtualization.” The paper, published earlier this year at SIGMOD 2025, details how Cockroach Labs built a system that offers the elasticity and cost-efficiency of serverless computing without sacrificing the performance and resilience of distributed SQL.

How can a multi-tenant architecture scale with you?

CockroachDB Basic and Standard are built on a multi-tenant, virtualized cluster architecture. Each tenant runs in its own isolated virtual cluster, allowing tenant workloads to scale independently while securely sharing underlying infrastructure.

_{Four differently sized virtualized clusters leveraging the resources of one CockroachDB physical cluster}

This design enables:

• Sub-second cold starts – resume from zero in under a second

• Automatic scale-to-zero – pay nothing when your cluster is idle

• Multi-region operation – run globally with distributed transactions

• Strong isolation – prevent noisy neighbors and ensure workload security

These capabilities make it possible for CockroachDB Cloud to offer low-cost, elastic database tiers that still meet the demands of modern, globally distributed applications.

From research to reality

While this paper was presented at SIGMOD 2025, one of the premier research conferences in databases, it represents technology that has been powering CockroachDB Cloud clusters in production for over four years. The paper dives into the core challenges and solutions that enable thousands of tenant clusters to operate efficiently, securely, and with predictable performance.

Read the Paper

Rebecca Taft is the Director of SQL Engineering at Cockroach Labs. During her 8 years at the company, she helped build the cost-based query optimizer from scratch and added support for features such as geospatial indexing and locality-optimized search in multi-region clusters. 

Prior to joining Cockroach Labs, she was a graduate student at MIT, where she worked with Professor Michael Stonebraker researching distributed database elasticity and multi-tenancy.  Becca holds a B.S. in Physics from Yale University and an M.S. and Ph.D. in Computer Science from MIT.