Join us for GraphQL Summit, October 10-12 in San Diego. Use promo code ODYSSEY for $400 off your pass.
Launch GraphOS Studio
You're viewing documentation for an upcoming version of this software. Switch to the latest stable version.

Caching in Apollo Kotlin


Apollo Kotlin supports multiple forms of client-side caching for different purposes. Its built-in normalized caches are especially useful for storing GraphQL results in an efficient, deduplicated way.

Additionally, Apollo Kotlin supports server-side caching of strings via persisted queries.

Here's a summary of all supported caches:

Built-in normalized cachesDeduplicates GraphQL response data, enabling your app to use it as a source of truth for your UI. Can also dynamically react to changes in that data. Both in-memory and SQLite caches are included, and you can use both to combine in-memory speed with SQLite persistence.
Built-in HTTP cacheCaches response data at the HTTP level. Can be faster to get started with and uses less CPU than normalized caches, but duplicates data and can't act as a source of truth.
External cachesIf you're already using an external cache (such as Room or SQLDelight), you can use it with Apollo Kotlin. Instead of using Apollo for both the network and database domains, you use generated models for the network and map them to your own database models.
Persisted queries (APQ)

With APQ, your client sends a hash of most queries to the server instead of the query itself. This makes requests much smaller and enables the use of GET requests instead of POST. This in turn makes it possible to cache results at the CDN level. The server also caches your query strings, meaning it doesn't have to parse a particular query each time it's sent.

Persisted queries can be automatic persisted queries (which require little configuration), or you can use custom persisted query IDs.

Client-side and server-side caches are not mutually exclusive, and most full-stack applications use both. Using a normalized cache with persisted queries helps you minimize your number of network requests, while also taking advantage of a CDN for requests that are required.

Persisted queries
Normalized caches
Edit on GitHubEditForumsDiscord