vscsiStats in ESXi


Interesting performance analysis method and ESXi 4.1 contains this tools within /sbin/vscsiStats (note the one Capital letter)

So enable ssh login on the host, login via SSH, run the tool.


  VscsiStats -- This tool controls vscsi data collection for virtual machine disk
                disk I/O workload characterization. Users can start and stop online data
                data collection for particular VMs as well as print out online histogram data.
                Command traces can also be collected and printed.

  The following histogram related options are available:
     -h, --help will print the usage
     -l, --list will list the available virtual machines and their virtual disks
     -r, --reset will reset the stats
     -s, --start will start vscsiStats collection; exclusive of -x
     -x, --stop will stop vscsiStats collection; exclusive of -s
     -w <worldgroup id>, --worldgroupid specifies a worldID to use for this operation
     -i <handle id>, --handleid specifies a vscsi handleID to use for this operation
           requires the -w option
     -p <histoType>, --printhistos will print out the current histograms for the specified
            histogram type. May be used in conjunction with -w and -i.
            histoType must be one and only one of:
                 all, ioLength, seekDistance, outstandingIOs, latency, interarrival
     -c, --csv will use comma as delimiter in conjunction with -p

  The following command trace related options are available:
     -t, --tracecmds will start scsi cmd traces; in conjunction with -s
         Note:- the -t option consumes significant system resources so
                enabling it indefinitely is not advisable
              - try to limit the #virtual disk for which cmd tracing is enabled at any
                given time by using --worldgroupid and/or --handleid options.
              - trace contains NO customer sensitive data
                - only information recorded is:
                  - serialnumber, IO block size, number of scatter-gather elements
                  - command type, block number, timestamp
                - Therefore, actual data/payload of commands is not stored
              - If successfully started, log channel id(s) will be printed out.
                To store the command trace in a file for later processing, invoke:
                $ logchannellogger <log_channel_id> <binary_trace_file>
     -e <trace file name>, --traceprettyprint reads in a vscsi cmd trace from the given
        filename and sends a CSV formatted output to stdout; exclusive of all other options

  vscsiStats Usage:
         vscsiStats [options]

Handy tidbit to collect performance data
vm-support -s -d <duriation to collect vscsistats snapshots> -i <interval between snapshots>
I Love it!

Additional Bits