Positive experience with usage of EHcache as replicated cache for WORM
What do you like best?
Ehcache solution could be easily integrated e.g. in you Spring Web solution, if you would like to to enable distributed caching. We have used it in one of our projects for WORM caching scenario (Write Once Read Many). Namely, we needed to implement asynchronous http communication with partner system for cluster environment. We have encountered the situation, when the response might come to another node than the one from which request has been sent. To solve this problem, we have enable distributed cache between nodes in the cluster based on eh-cache.
What do you dislike?
In our particular situation it was hard to find out, what was the reason of missing synchronization between nodes. The reason was setting of 1s for default synchronization interval between nodes. This setting was also poor documented (RMICacheReplicatorFactory.asynchronousReplicationIntervalMillis). Finally we have reconfigured this setting to shorter one (10 ms) after some failures in acceptance test environment and extensive analyze.
Recommendations to others considering the product:
Thoroughly read documentation and properly set up parameters.
What problems are you solving with the product? What benefits have you realized?
Replication between nodes in clustered application for asynchronous http communication with partner system. Easy integration with Spring.