Eric's Technical Outlet

Learning the hard way so you don't have to

SCVMM Maximum disk I/O per second setting

There is a bug in SCVMM, both RTM and SP1, that doesn’t properly handle the possible values for this field. That was fixed in a later rollup. The default value is still too low for most installations.

When trying to LiveMigrate a VM in SCVMM, it may give the target host a very low rating because “The projected disk I/O utilization (IOPS) x exceeds the maximum disk I/O utilization of y.” By default, y is set to 1611. You increase this by accessing the properties of the host and/or its host group in SCVMM. The “Reserves” tab contains the field that controls this.

The placement is somewhat misleading, as this isn’t actually a reserved resource like the others on the page. In this case, if moving a VM to this host would cause its anticipated I/O load to exceed the displayed number, it will generate the rating reason displayed above. Of course, it is just a projection based on recent activity; running Windows Update on a minimally used VM and then trying to LiveMigrate it could cause this message.

This field is of minimal usage for many installations. Your hosts can almost undoubtedly handle more IOPS than your shared storage can. If you have more than one shared device accessible to your cluster, then there’s no way for a single number to determine how a host should operate. If you do only have one shared storage location that all the hosts access, then it doesn’t really matter which host the VMs are on, because ALL of their I/O is ultimately from the same unit. Probably the only configuration where this will be useful is when each host has access to such a large number of shared storage locations that it would actually be conceivable that a host could be overwhelmed.

To all but guarantee that the message is never displayed, set it to a high number, like 10,000. On earlier editions of SCVMM, if the number was too high, it would roll it over and represent it internally as a negative number. If you want to put a more accurate number in the field, then your storage vendor will be the best point of contact to determine what sort of IOPS their equipment handles.

Note that you can always use Failover Cluster Manager to LiveMigrate a VM when SCVMM refuses to move it due to IOPS.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: