Cache behavior and performance

Guidewire generally distinguishes two types of caches:

Server cache – A cache that is purely local to the PolicyCenter server
Database cache – A cache used by a database to hold data retrieved from storage

Proper database caching behavior is critical to performance.

The server cache is purely local to each individual PolicyCenter server. Therefore, one server instance might contain information on a specific object while another server might not contain that information.

For example, if a PolicyCenter user works on a policy, PolicyCenter loads corresponding objects on the server to which the user connects. If another user must approve the action of the first user, the approver user might interact with another PolicyCenter server. In that case, the second server likely does not have the corresponding information in cache. Therefore, the approver might experience slower performance as the server must populate the cache.

Cache content is lost every time that you stop a PolicyCenter server. Therefore, after you start a PolicyCenter server, expect lower performance during a ramp-up phase.

Batch processes leverage the cache mechanism. Batch jobs can work on many objects. Therefore, batch jobs can use the cache extensively. This can have the adverse effect of prematurely evicting objects from the cache, thereby forcing additional cache loads. If you run many intensive batch processes, Guidewire recommends that you have a dedicated server with the batch role with no online traffic directed to it.