Skip to main content

Overview

Momento Cache is a fully-managed, dedicated Valkey cluster. Provision isolated capacity operated by Momento, then connect with any Valkey or Redis client through a managed gateway. An HTTP API for RESP commands is planned.

Unlike a multi-tenant serverless cache, Momento Cache gives you a dedicated backend sized to your workload, with isolation at the capacity pool boundary. Unlike running Valkey on your own infrastructure, you do not manage nodes, failover, or upgrades.

Variants​

Momento Cache comes in two variants that differ only in how you size capacity:

  • Cluster: configure instance type, shard count, and replica count
  • Flex: configure storage size in GB and select a performance profile

Both run the same Valkey engine, the same gateway, and the same isolation model, so the rest of this documentation applies to both. Documentation for the Serverless variant of Momento Cache is available here.

When to use it​

Use Momento Cache when you want managed Valkey with dedicated, predictable capacity: a replacement for ElastiCache, MemoryDB, or Memorystore, or managed Valkey on a cloud where you run today. You get standard Valkey compatibility and multi-AZ resilience without operating the cluster yourself.

The model​

Momento Cache has two primitives:

  • A Capacity Pool is the dedicated Valkey capacity you provision. You size it (by instance for Cluster, or by capacity for Flex) and choose its availability zones; Momento provisions and operates the underlying cluster.
  • A Database is a logical cache that runs on a Capacity Pool. You can run several Databases on one pool; they share its compute and memory.

You reach a Database through a shared gateway using a standard Valkey client. Isolation is guaranteed at the pool boundary.

Get started​

The quickstart walks through provisioning a Capacity Pool, creating a Database, and connecting a client.

For the control-plane API, see the reference.

Availability​

Momento Cache runs on AWS today. Support for additional clouds is planned.