Measuring the Datacentre

The big push in Datacentre Capacity Management currently seems to be related to the power necessary to keep the infrastructure operational and to keep it cool.

Many suppliers are pushing their "Green IT" products and boasting about how little power is consumed.  While this is undoubtedly a good thing I wonder if it is leading people to forget the basic feature of a datacentre; Space.  If your Datacentre provision runs out of power you can purchase additional generating capacity from your provider, and hopefully upgrade your UPS so that you can still enact your BCP in the event of a failure.  If you run out of cooling capacity, you could even decide to run your datacentres a little "hot" for a while, or alternatively, spend money on buying more capacity.  Both of those upgrades (Power and Cooling) can be put in place in a relatively short period of time.  However, if you run out of space in your datacentre, then the only things you can do is to buy another datacentre, or build an extension.

Both of those actions take a considerable amount of time.

Blade computing has reduced the amount of space that is required for a "server", and virtualisation has reduced this even more.  But a blade enclosure still takes up space, and a virtual guest still relies on a physical host on which it can reside.  The issue of Space management is just as crucial as it ever was.

We are currently working with a client to try and revamp their datacentre reporting.  They have been trying to implement a single value that indicates the amount of space remaining in each of their datacentre suites in terms of “racks”.  The issue that they have experienced, is that the assumptions they have made regarding the amount of power and cooling that a rack requires has led to the following quirk: Some of the datacentre suites have a negative number of racks free.

How can this be?  A quick walk around the datacentre shows that there is plenty of free space, there aren’t racks piled up on to of each other.  Similarly there appears to be enough power to keep everything switched on, the UPSs aren’t beginning to whine and melt.  Neither are the operators beginning to “whine” or “melt” due to the tropical temperatures one might be expecting.  So why are the reports showing negative space free?  The maths behind the reports is valid, so all that remains is that there must be an error in the logic being used.

This will need further investigation.

This entry was posted in Blog Posts, Capacity Management and tagged , , , , , , , , . Bookmark the permalink.