One of the most widely used monitors in Robot Monitor is the built-in Disks Busy monitor, which shows you how busy your storage system is.
Ever wondered how many of the I/O requests the operating system (the licensed internal code or LIC, actually) sent to the disk since the last measurement interval were serviced immediately? How about the number of requests that had to wait?
Disks Busy tells you the percentage of those requests that had to stand in a line and wait. It’s the best indication of the I/O load on your disk system.
Guess what? I want you to stop using it…if you’re using more than one auxiliary storage pool (ASP).
Enter, the ASP Busy Monitor
This may come as a surprise to you, but I can assure you there’s a good reason: Disks Busy works across all physical disks that are used on your system—that means across all ASPs.
You may have been thinking that the Disks Busy figure is only for your System ASP. Think again! If you have more than one ASP, Disks Busy is not just reporting on your System ASP. Rather, it reports the average figure for all of them.
If you have nine independent ASPs and your System ASP is being crushed under the weight of a complete, full, utter maximum I/O load, Disks Busy will barely see 10 percent. (Assuming the IASPs are idling at that time and that all APS have the same number of physical disks.) So, the more ASPs you have, the more the Disks Busy information for any individual ASP will be diluted.
Here’s what you should do instead:
- Add new ASP Busy monitors, one per ASP. ASP Busy is like Disks Busy, but it only returns the values for one ASP.
- Display the newly-created monitors.
- Assign thresholds to the monitors. (You can use the Disks Busy thresholds, too.)
- Use ASP Busy to report on the individual ASPs rather than Disks Busy.
- Upgrade to the latest version of Robot Monitor.