At the cluster level, the performance monitoring service shows performance metrics for virtual machines running on vSAN as well as the vSAN back end. These metrics provide quick visibility to the entire vSAN cluster, showing how vSAN and the objects that reside on it are performing.

The following figure shows metrics such as IOPS, throughput, latency, and outstanding I/O not only for virtual disks, but for all vSAN objects in the cluster.

 

Back end vSAN metrics help show what is required to deliver the expected performance, and what is visible at the virtual machine and object level. It is noteworthy that the I/O overhead from different storage policies can show significantly different back end metrics than what is observed at the virtual machine or object level.

Take the example of a virtual machine with a Number of Failures to Tolerate of 1, with a failure tolerance method of RAID-1 (Mirroring). For every write I/O to a virtual disk, two are seen on the back end. This is because two mirrors both receive a write, despite the fact that the virtual disk only had a single write. Changing the fault tolerance method to RAID-5/6 (erasure coding) now requires four writes, because data and parity in a RAID-5 configuration is comprised of three data writes and one parity write.