Opserver
Opserver copied to clipboard
Proposal: Change Disk Performance reporting
Currently the the new disk/volume performance reporting is inconsistent between providers. For example for WMI it's pulling the number of bytes Read/Write whereas the Orion provider is giving the number of reads/writes.
The number of bytes being read/written to disk only tell half the story as that differs wildly between for example DB servers and web servers and is really dictated by the type of disks (NAS volume/HDD/SSD), and a pure throughput number doesn't actually tell you if it is affecting performance.
My proposal is that it be changed to rather report on the number of reads/writes, and also adding disk queue length and disk latency metrics as those give in my opinion a much clearer image of the performance impact disk/volume IO might be having on the server. These metrics are also properly reported in all providers.
Discuss! :)