Disk Space Monitoring–Common Practices

If you have been monitoring disk space with SCOM for while, you are very likely aware of how the default disk space monitor works. It has been discussed here by Kevin Holman as well. In a nutshell, before the 6.0.6972.0 version of the core OS MP, there was only one “Double threshold” monitor. I like to call it “The short blanket” monitor, since when you cover you head, you expose your feet, and vice-versa.

With 6.0.6972.0 , Microsoft broke it into two separate monitors, one for Megabytes and one for Percentage monitoring. I really like the idea. However, even with that, in a real world scenario, you can still have volumes that vary from 50Gb to Terabytes of space. A percentage monitor usually sounds like a good idea, but when you think of 10% of 50Gb and 10% of 2Tbs, and you get an alert at 2AM because your disk has only 200Gb free, let’s say it could be avoided.

Generally, the Mbytes options has been the more common. Even though you can still have some distortion and some possible issues, it will be more of an absolute number to deal with. An example:

image

Now, there is a situation where you may get burned, to consider just one case: what if the type of data written to a certain volume can grow very fast, for example, large media files are written quickly or even VM disks, or something the like. The 5000Mb can go down too quickly and you won’t have time to act. You can of course increase the limit to 100000Mb (100Gb), but for many other disks, it will warn you too early.

But how to be selective? You can always override here and there, for specific disks, but it can be a lot of overrides to manage (maybe there are very special situations where you have to do it anyways). However, my general recommendation to customers is:

0. If you can live with some distortions here and there, ok, create a new general and broad override for the logical disk classes and you are good. If not:

1. Create dynamic disk groups, based on the volume sizes.

I first thought of creating a new discovery,etc. Not needed. You can leverage what SCOM already provides! Here’s how:

Go the the Authoring area and Create a new Group:

image

image

Skip the explicit members and create a membership rule by selecting Logical Disk (Server) and then click Add.

image

Select Size (Mbytes) and set your limit (50Gb in my case):

image

Once created, verify that the group was populated correctly:

image

image

(You may also want to filter the Mounted Disks or very small disks, whatever suits you better).

Having the groups set, you can then create specific overrides for the different categories, as below:

Go to the Authoring->Monitoring Section:

image

Change your scope to make it easier to find:

image

Make sure you disable the original Logical Disk monitor:

imageimage

And enable the new MB only monitor:

image

image

Repeat the process to Windows 2003 and Windows 2012 Logical disk monitors.

Now set the threshold for your groups (yes, for 2003 and 2012 as well…I know…):

image

image

image

In this case, you may care or not whether the disks are system disks or non-system disks, but it is essentially up to you.

Create groups and overrides for the different categories, maybe medium sized disks, very large disks,etc and you’re good.