also are there any metrics that show what percentage of the maximum cache size is being used? my theory is that our requests only fire against the dispatch cache, which is smaller, thus many entries get evicted again.