Has there been any progress on addressing the long standing performance issues with messaging?
i.e. messaging stores a queue within a single cache entry, which means it needs to deserialize, add item, and reserialize every time we add an item to the queue.
For push rep, this means a burst of data can bring messaging to it knees and cause a cluster to fall over (eg: a clear of a large cache, or a remote site that is unavailable causing queues to grow to a very large size).
I have also noticed that when a queue gets very large, the jmx suspend/drain times out, and throws an exception.