Storage contention guidance

The following table provides general guidance on storage contention. It must be read in conjunction with the CPU Contention Guidance section in Chapter 4, CPU Counters. I must emphasize that these are just guidelines—you should set the values according to the unique needs of your environment.

Object	Contention	Remarks
VM	Official SLA: Tier 1: 10 milliseconds Tier 2: 20 milliseconds Tier 3: 30 milliseconds Internal threshold: Tier 1: 5 milliseconds Tier 2: 15 milliseconds Tier 3: 20 milliseconds	The internal threshold is lower as it serves as a buffer.

Your customers only care about their own VMs, so the following numbers are for your internal targets. You should have a more stringent internal threshold. This gives you a buffer to troubleshoot.

Object	Contention	Remarks
ESXi	NA	There is no need to track at this level because of the following reasons: You already track contention at the VM level, datastore level, and array level. VMs get vMotion-ed in the cluster...