Skip to main content

Computes and endpoints

info

Lakebase Autoscaling is the latest version of Lakebase, with autoscaling compute, scale-to-zero, branching, and instant restore. For supported regions, see Region availability. If you are a Lakebase Provisioned user, see Lakebase Provisioned.

In Lakebase, you connect to your database through a Lakebase endpoint, a stable database access point identified by a UID. Behind the endpoint, one or more compute instances provide the processing power to run your queries. Your connection strings stay the same even as you scale compute resources or enable high availability.

A branch typically has one read-write endpoint (your primary compute) and optionally one or more read-only endpoints (read replicas):

Branch: production

├── Endpoint: primary (ep-rapid-cherry-y12fhu7x)
│ └── Compute instance (read-write)

└── Endpoint: read-replica-1 (ep-cool-breeze-a8xk2m3p)
└── Compute instance (read-only)

With high availability, a single endpoint contains multiple compute instances. Your connection string does not change:

Branch: production

├── Endpoint: primary (ep-rapid-cherry-y12fhu7x)
│ ├── Compute instance (read-write primary)
│ ├── Compute instance (secondary, failover)
│ └── Compute instance (secondary, failover)

└── Endpoint: read-replica-1 (ep-cool-breeze-a8xk2m3p)
└── Compute instance (read-only)

The UI refers to these resources as computes and read replicas. The API, SDKs, and CLI call them endpoints, with an endpoint_type field that indicates the role:

UI term

API endpoint_type

Primary compute

ENDPOINT_TYPE_READ_WRITE

Read replica

ENDPOINT_TYPE_READ_ONLY

Next steps