I have set up a hbase cluster. The cluster consists of 5 servers each with 32G ram and 8 cores. Hbase fails are increasing whenever there is a spike in disk latency. I can't figure out from graphs why the spikes are coming on the disks. I am attaching the related graphs.