Caches should be replicated asynchronously to improve app scalability

Caches should be replicated asynchronously to improve app scalability.

Synchronous replication may cause problems with scalability, as each change in cache needs to be replicated to other nodes. It means each change is blocking other actions for the caller until completed. Second important thing is that replication event is not guaranteed to be delivered. If something goes wrong, there is an exception when replicating - the operation will not be redone and cluster may be inconsistent, until cache will be refreshed.

Regarding (Performance and scalability) the question there is not about performance but rather scalability. Please take a look on actions 9, 11, 32. They take much more time on DC 2 nodes than on DC 1 node (or standalone I'm not sure about configuration). This might indicate that e.g.:

• there is a scalability problem e.g. cache replication is synchronous and takes a long time
• measurements results just have high variance
• the environment with DC 2 nodes was not the same as for DC 1 node
  Please investigate if the app doesn't have scalability problem.

See https://docs.atlassian.com/atlassian-cache-api/2.5.0/atlassian-cache-api/apidocs/com/atlassian/cache/CacheSettingsBuilder.html#replicateSynchronously().

The feedback was collected from:

• https://ecosystem.atlassian.net/browse/DCHELP-131
• https://jira.atlassian.com/browse/JRASERVER-67019