PgCache vs Amazon ElastiCache

Amazon ElastiCache is AWS's managed cache service. It runs Valkey, Redis OSS or Memcached. To cache RDS or Aurora reads in ElastiCache, your app writes query results into the cache and clears them when the data changes. PgCache sits in front of RDS or Aurora PostgreSQL, picks what to cache on its own, and keeps the cached data current through replication.

See which of your own queries PgCache would serve: check your workload or open a sample report.

ElastiCache facts checked against AWS's docs on October 15, 2026.

PgCache

A caching proxy that runs in your VPC in front of RDS or Aurora PostgreSQL. It stores the data for your repeated reads in a separate, vanilla Postgres instance and keeps that data current through logical replication.

No cache code or TTLs in the app
Writes from any app, job or tool reach the cache
Each connection reads its own writes

ElastiCache

AWS’s managed cache service, running Valkey, Redis OSS or Memcached. Your app reads and writes the cache by key.

Where they differ

Topic	PgCache	ElastiCache
App changes	Point the connection string at PgCache	Cache code around your queries
Freshness	Within typical logical replication lag	Stale until your code clears the key or its TTL expires
Writes from other services and tools	Reach the cache through replication	Missed unless each one also clears the cache
What you run	PgCache on EC2 or in a container	A managed cache cluster, plus cache code in your app

Which one fits

Choose PgCache when

You run RDS or Aurora PostgreSQL 16 or later
You don’t want cache code or TTLs in the app

Choose ElastiCache when

You cache sessions, API responses or other data that isn’t a query result
Your database is MySQL, or Postgres older than 16

What they have in common

Both take repeated reads off RDS or Aurora
Both run inside your VPC

Using them together

ElastiCache for sessions and other data that doesn’t live in Postgres
PgCache for the SQL reads you’d otherwise cache by hand

Common questions

Does PgCache use ElastiCache?

No. PgCache stores cached data in its own Postgres instance, so there’s no separate cache cluster to run.

What about AWS's JDBC query cache plugin?

AWS’s JDBC wrapper has a plugin that caches query results in ElastiCache for Java apps. You add a TTL hint to each query you want cached. PgCache needs no hints, works with any language, and keeps cached results current through replication.