On clusters with running CQs, a performance degradation was introduced by
GEODE-6989 for regular cache operations.
Every time an entry is updated on the cache, Geode internally (and synchronously) executes the registered CQs on the event entry to determine whether the subscribed clients should be notified or not. Due to the changes introduced through
GEODE-6989, these regular cache operations now takes longer than expected (build internal cache key, check internal cache, and subsequent put if needed), reducing the throughput ~ %10.