[ https://issues.jboss.org/browse/ISPN-3792?page=com.atlassian.jira.plugin.... ] Karl von Randow commented on ISPN-3792: --------------------------------------- [~dan.berindei] Thanks for your comment. I use an invalidation cache across my cluster, and I use it for caching fragments of HTML where some of those fragments have different retention characteristics. I specify the retention characteristics with the fragment, and I have lots of such fragments, so rather than making an individual cache for each of them I would like to use one cache and simply vary the metadata. I need to use PFER because it's an invalidation cache. I think PFER is more essential than just Hibernate 2LC; I think it's essential for any invalidation cache. I can absolutely see your point about more methods in the Cache interface. If it were possible to use a flag to achieve the same result, I would. I think the only thing not achievable by flags is the transaction suspend - which I don't need, but I agree with the sentiment of it, and some of my caches are transactional. I would be completely happy with just the method in AdvancedCache. I do think it's worth considering whether PFER is more important than that when using invalidation though. If you don't use it you pretty quickly get into a dysfunctional cache scenario with each cluster member invalidating the other. It seems that invalidation caches are more likely than not external read caches! -- This message was sent by Atlassian JIRA (v6.3.1#6329)