Cockroach Plenum Overview

Cockroach Plenum is Cockroach Continuum’s elastic storage layer that separates a cluster’s storage from its compute. Plenum provides a custom storage architecture purpose-built for CockroachDB that preserves CockroachDB’s existing while removing reliance on a provisioned disk layer. On Continuum, Plenum backs the usage-based storage model, available on eligible clusters. Usage-based storage is billed based on physical-used GiB rather than provisioned capacity.

​

How it works

CockroachDB nodes write to Plenum through the instead of writing to attached disks. Plenum manages the physical placement and redundancy of stored data across availability zones. Above the storage-layer boundary, , , and behavior remains unchanged.

​

Compared to provisioned storage

With provisioned block storage, each node has its own attached volume, so storage and compute are sized and scaled together as a unit. You provision capacity for the data you expect, pay for it whether or not it is full, and resize it as data grows. Provisioned block-storage services such as Amazon EBS also replicate data internally, adding cost and redundancy on top of CockroachDB’s . Plenum removes the provisioned-disk layer entirely. CockroachDB nodes write to a shared storage service instead of to attached disks, and storage is billed based on physical-used GiB.

​

Usage-based storage

When you , you choose its storage model:

• Usage-based. Storage is backed by Plenum. You do not provision storage capacity or bandwidth. Storage capacity scales automatically as data is added or removed. You are billed based on physical-used GiB, and the Console shows storage consumption rather than provisioned capacity.
• Provisioned. Storage is backed by conventional block volumes and billed as provisioned capacity.

Usage-based storage is available on single-region Mission Critical clusters running CockroachDB v26.3 or later. Multi-region Mission Critical clusters and host clusters use provisioned storage. Plenum is initially available in the following AWS regions:

• us-east-1
• us-west-2
• eu-central-1
• eu-west-1

Standard clusters run on shared infrastructure with usage-based storage billing, separate from Plenum. The storage model is chosen when you , and cannot be changed. To change storage models, create a new cluster and move the workload. For details on how storage is billed, refer to .

​

Independent scaling

Because the nodes running CockroachDB are stateless and separated from the storage layer, storage now scales independently.

​

Durability and security

Plenum provides redundant storage across availability zones in a region. Data is encrypted at rest with its own keys, so clusters sharing Plenum’s storage service are cryptographically isolated from one another.

​

Demo video

For a demonstration of how to use Cockroach Plenum and cluster virtualization, watch the following video:

​

See also