A cluster/group is formed with 2 nodes with Cache Mode = REPL_ASYNC. We would typically have around 20 updates/sec in the Cache.
If we use the cluster/group, the performance of the application goes down drastically. If we disable the cluster/group, then the application performs well.
This combo is from the stone age; I strongly recommend you upgrade to a more recent version.
In our HTTP session replication tests, we got numbers of thousands of updates / node / second, so 20 updates is peanuts...
Could be a config issue...