Amazon ElastiCache

Amazon ElastiCache is a managed in-memory cache from AWS for Valkey, Redis OSS and Memcached, run serverless or on nodes you choose and billed by usage.

ElastiCache keeps the data an application reads most often in memory, so requests are answered in microseconds instead of waiting on a disk-based database. Amazon Web Services builds and runs it as a fully managed service inside the AWS cloud. Caches speak the open source Valkey, Redis OSS and Memcached protocols, which lets existing client code connect with few or no changes. Each cache runs in one of two modes: ElastiCache Serverless, with no infrastructure to manage and no maintenance windows, or node-based clusters built from chosen node types. Valkey and Redis OSS caches hold strings, hashes, lists, sets, sorted sets and streams, with expiry times and eviction policies for when memory fills. Cluster mode enabled spreads the keyspace across shards, and automatic failover promotes a read replica when a primary fails. Backups land in Amazon S3, where they can restore a cache or seed a new one. On recent Valkey versions, node-based clusters add durability, which writes every acknowledged write to a Multi-AZ transactional log. The same clusters offer vector, full-text, numeric and tag search over hashes and JSON documents. Data tiering moves less-used data from memory to local SSD on a dedicated node family, and Global Datastore copies a cache to other AWS Regions for local reads. The service is built for developers and platform teams whose applications already run on AWS and need faster reads than their primary database gives. Teams running their own Redis OSS, Valkey or Memcached servers can move to it with little or no change to application code. Builders of generative AI and agentic applications use semantic and prompt caching to serve repeated or similar queries without calling the model again. Serverless suits teams that would rather not size clusters at all, while node-based clusters suit teams that want to choose node types and counts for a steady workload. Role-based access control scopes each user to chosen commands and key patterns, alongside AWS IAM and Amazon VPC isolation. A web application keeps user sessions, shopping carts and product catalog entries in a cache so any server can pick up the next request. Leaderboards run on sorted sets that rank entries as scores change, and pub/sub channels fan messages out to every subscribed client. A team leaving a self-managed cache brings its data across with AWS Database Migration Service, then watches the cache in Amazon CloudWatch. A multi-Region application reads from a Global Datastore copy close to its users while writes go to one primary Region. With durability turned on, a Valkey cluster can also act as the primary data store for records that must survive a failure.