Allthough I'm not familiar enough with the code to understand every implication of the proposed implementation, I think that the idea is sound. I encountered blocking reads in our application which need to be addressed. In the end I constructed a local cache in front of the distributed cache. The local cache is updated by the distributed in a fashion where we never block readers. (Optimistically locking was not a perfect fit here, I might revise that in the future if I get the time). The blocking encountered was not only due to write locked nodes but also synchronization blocks. However, the local cache solved the issues and I now get very performant concurrent reading from the cache layers. If I haven't gotten this all backwards this is a bit similar to what you propose. And the implications by this is that you would infact be solving a real problem that at least we had to deal with. I think that given the application areas for a cache, fast, non-blocking read access is definitely crucial. And I would love to see it implemented in the core cache instead of a tacked on solution like mine. =) View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4066397#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...