I am starting to gather some disk statistics with get-stat. I am curious the differences in the performance counters of: disk, virtual disk, and datastore.
I am assuming they deal with the different virtualization layers going all the way through the kernel to the actualy lun.
However I cant find the exact deliniation on exactly what layer each of this includes. If someone has the exact line, could they post. Thank you.
That is correct, there are metrics for all the layers of storage (vdisk, datastore,LUN...).
All the performance available metrics are listed in the PerformanceManager object in the SDK Reference.
For more details on the statistics have a look at my statistics series, starting with PowerCLI & vSphere statistics – Part 1 – The basics
The post also includes a number of links to other sources of information.
Some of the datastore metrics are not readily available through PowerCLI, see my Datastore usage statistics post.
Blog: lucd.info Twitter: @LucD22 Co-author PowerCLI Reference
Hi,
Kindly have this pdf. Its is having very good infomatory as per our query
http://www.vmware.com/files/pdf/vmfs_rdm_perf.pdf
That is correct, there are metrics for all the layers of storage (vdisk, datastore,LUN...).
All the performance available metrics are listed in the PerformanceManager object in the SDK Reference.
For more details on the statistics have a look at my statistics series, starting with PowerCLI & vSphere statistics – Part 1 – The basics
The post also includes a number of links to other sources of information.
Some of the datastore metrics are not readily available through PowerCLI, see my Datastore usage statistics post.
Blog: lucd.info Twitter: @LucD22 Co-author PowerCLI Reference
Thank you both of you. One question is I cannot distinguish between "disk" and "datastore". "virtual disk" i know is before the question even gets to the hba and I assume will measure some of the kernel latency. But how are disk and datastore distinguished?
I also assume to get detailed round-trip numbers i will have to get the storage adapter involved in some way--but I would think that would be in disk and datastore lateny more or less.
The disk metrics are two-fold, for a hostsystem they show IO metrics to/from the LUN, for a VM they show IO metrics to/from a vdisk.
The introduction of each group under Storage IO on the PerformanceManager page gives more info on what the metrics in a specific group are about.
Blog: lucd.info Twitter: @LucD22 Co-author PowerCLI Reference