Hello,

We are going to use Hibernate on top of an existing REPLICATED database.

This means that there will be data beeing added, changed and removed behind Hibernates back at least every night, sometimes also during ad hoc replication.

This all works out nice, but I think we will have issues related to caching.

When replicating this way, do we need to refresh any caches? If so, is there a way to do so programmatically for all caches?

Not sure if we should have a thread that does this periodically and programmatically... or are there better ways?

The solution must work on all kinds of database products (which is why we use Hibernate to begin with :)

Ideas are appreciated!

In most caches you can specify a time to live for each region. Won't that do enough for you?



Farzad=

No, because the replication can occur anytime when running it ad-hoc (i.e. after some server has been down)

So I think I really need some way to force an update of the caches.

this might not be the best solution but worst comes to worst you can access the underlying cache technology classes. For example, ehcache can export some mbeans. I am not aware what kind of things you can do through them but it is very likely you can manipulate cache regions. In any events, how do you programmatically realize it is time to evict a cache region? Isn't it better to set up a short life span cache region for sensitive data? This is not horrible, depending on uses cases, since db engines usually have a fairly strong cache.



Farzad-