If an application modifies an item whose item descriptor specifies distributed TCP caching mode, a cache invalidation event is broadcast from that Oracle ATG Web Commerce instance to all other Oracle ATG Web Commerce instances that use distributed TCP caching. The event message supplies the nature of the change, the changed item’s type, and repository ID. Receiving repositories respond by invalidating that cached item.

Distributed TCP caching is suitable for sites where the following conditions are true:

An item that changes frequently—say, more than 50 or 100 times per second —is not suitable for distributed TCP caching mode, because the extra network activity incurred by cache invalidation messages outweighs caching benefits. Cache invalidation events are broadcast to all Oracle ATG Web Commerce instances that enable distributed caching, even if they do not cache the invalidated item; as the number of these Oracle ATG Web Commerce instances increases, so too increases the network activity associated with each cache invalidation event.


Distributed TCP caching is implemented by the following elements:

The following graphic shows how updates to a cached repository item on Oracle ATG Web Commerce server Alamo trigger an invalidation event that is broadcast to other Oracle ATG Web Commerce servers, whose GSAEventServers relay it to the corresponding repositories: