Cockroach Continuum: The Agentic Database Cloud by Cockroach Labs

Idle capacity is now too costly to waste

Treating every database as a fixed island multiplies idle capacity and operational overhead.

90%

of clusters operate at less than 20% CPU usage

85%

of clusters use less than half the storage they provisioned.

29%

disk utilization across an entire cloud fleet

^{Source: Cockroach Labs analysis of fleet-wide production telemetry measured in July 30, 2026.}

Agent scale without compromise

Keep the correctness, resilience, and scale you rely on, and gain the control, and simplicity that growth usually costs you.

CLUSTER VIRTUALIZATION

Isolated databases. Shared capacity.

Give every workload its own isolated database while sharing capacity across the estate, or use dedicated capacity when mission-critical workloads require it.

COCKROACH PLENUM, USAGE-BASED STORAGE

Pay for bytes, not headroom.

Separate storage from compute so cost follows the data you hold, not peak capacity you reserve.

COCKROACH AEGIS, AGENT-ASSISTED OPERATIONS

It finds the problem. You approve the fix.

Aegis observes and diagnoses across the estate, then recommends what to do next. You approve every change.

Give every workload isolation. Operate the estate as one.

See how cluster virtualization, Cockroach Plenum, and Cockroach Aegis work together across shared and dedicated workload models.

COCKROACH CONTINUUM PRODUCT DEMO

See the estate move as one system

Watch the estate respond as demand changes: virtual clusters share capacity across workloads, Plenum separates compute from storage, and Aegis observes, diagnoses, and recommends the next move for approval.

Watch video

CockroachDB is the foundation customers already trust

82K+managed clusters

CockroachDB powers 82K+ managed clusters on CRDB Cloud, giving Continuum a production foundation for a larger estate.

BUILT ON COCKROACHDB

Continuum is new. The foundation is proven.

Cockroach Continuum is built on CockroachDB, proven where correctness and resilience is non-negotiable.

A larger estate without a larger operating burden.

Continuum brings estate-scale capabilities to the CockroachDB foundation customers already trust, with the controls and proof points to evaluate the path forward.

Everything your database needs. Nothing more, nothing less.

Resilience across zones and regions

Survive infrastructure failure without a failover. If agents are down, your business is too.

Correctness under concurrency

Serializable isolation by default. Thousands of agents writing at once can't reason about weaker guarantees.

Petabyte scale without downtime

Estates that grow this fast can't pause to resize.

Shared or dedicated, per workload

Mission critical and high density are different jobs. Match the infrastructure to the isolation and availability each one needs.

Estate scale elasticity

Tens of thousands of databases sharing capacity, created in milliseconds, billed on use.

FAQ

Common questions about Cockroach Continuum.

Cockroach Plenum is Cockroach Continuum's usage-based storage: it scales on its own, and you pay for what you store, not what you provision.

Cluster virtualization lets you run many isolated databases on shared infrastructure, so each app gets its own database without standing up separate clusters.

Cockroach Aegis is our AI operator for CockroachDB: it watches your databases, flags issues, and recommends fixes rather than acting on its own. Aegis works with all CockroachDB deployments.

Cockroach Continuum offers both, and capacity can be shared in two ways. Shared infrastructure (Standard) is multi-tenant: your databases run on capacity Continuum shares across customers, so you get a database without paying for dedicated hardware. Dedicated infrastructure (Mission Critical) is provisioned for you alone, for workloads that need mission-critical isolation or availability. And with cluster virtualization, you can run many of your own isolated databases on a single dedicated host, giving each app its own database without a separate cluster. In every case, each database stays logically isolated.

Know your estate before you scale it.

The estate assessment maps your current database footprint against Continuum's model: cluster count, workload distribution, agent readiness, and what virtualization would actually cost you. No commitment. One conversation.

Try For Free

Talk To Our Team

Operate a dramatically larger database estate

300+ database nodes

82K+managed clusters

1.2M writes per second

300+ database nodes

82K+managed clusters

1.2M writes per second

300+ database nodes

Continuum is new. The foundation is proven.