APIGEE Cache Eviction Policy

[sagrawal]

Team, 

In my project we want to cache api responses but we have millions of variations which will exhaust the cache size (2.5 million on L2 Cloud Cache). I could not find in document what will happen if cache size gets exhausted, how cache eviction will happen? is it based on FIFO, LIFO or LRU (lease recently used). is there any control we have to define this cache eviction policy?

Another question is if we want to cache say 10 million responses then how to distribute caches into multiple fragments to maintain such a large cache. 

[dknezic]

Should be evicted based on LRU and this is not configurable.

This doc has more details https://docs.apigee.com/api-platform/cache/cache-internals

[dchiesa1]

If caching is one of your main purposes, and if you have o(10 million) of records to cache, which is above the limit that can be cached on Apigee per the limits documentation, then you may want to consider building your own caching layer on Redis or something similar, and using a microservice to manage that cache.