Graphing CMDS/s / IOPS from VMWare ESX hosts

vmware-esxivmware-vsphere

If I look at esxtop in "disk adapter" view I can see CMDS/s.

How can I extract this information in a counter like way so I can graph it?

This doesn't need to be done in real time, I'd be happy to log the data on all hosts and then extract it later to analyse it.

Best Answer

You mean, apart from using the pretty graphs in vCenter?

The simplest option is to run the command

# esxtop -b > esxtop_whatever.csv

This will make esxtop run in "batch mode", outputting data into CSV format. (See also: http://www.jume.nl/esx4man/man1/esxtop.1.html)

You can then load this file into perfmon on Windows or use esxplot (http://labs.vmware.com/flings/esxplot) to view the data.

Another nice trick is to create a performance snapshot (http://kb.vmware.com/kb/1967)

# vm-support -s -d <duration> -i <interval>

This will create a large file with a number of 'snapshots' of the system state. This can then be 'replayed' with the command

# esxtop -R .

in the extracted TGZ file that was created in the previous step.

We use that one quite a bit for performance troubleshooting at VMware ;-)

Related Topic