Where do we go from here?

People often ask me what I think about the quest to develop the 'productivity metric'. I often pause before answering that and then ask 'do you mean DCeP, CUPS, CADE, DPPE or one of the many proposed proxies?'

Normally I get at that point that implies the fact that the person that asked the question thinks I'm a little too smug for my own good! However, I'm not trying to be smug, I'm simply trying to illustrate that 'productivity' is so poorly defined that we have a whole number of competing metrics claiming to have a good handle on the issue.

From my perspective the whole notion of boiling down (or up depending on which way you look at it) everything going on in a data centre into a single productivity number is meaningless unless you happen to be Google or someone like them and can hand on heart say that pretty much your entire data centre is carrying out one or perhaps two unique tasks.

In enterprise land, where most of us live such a position is just plain silly. My data centre does all sorts of things all of which contribute to the overall productivity of the data centre but cannot simply be amassed together and represented by a single number of dimensionless units.

Some say it's like PUE, you can't use it to compare two different sites but you can use it to compare the same site over time. Well can you do that with these productivity metrics? Well I'm afraid not. Not in any meaningful way at least.

So what are we going to do? How can we describe the value of what the data centre produced, how do we benchmark, compare and optimise, what data centre level yardstick can we apply?

There is only one group that can tell us the value of what the data centre did...how productive it was, and that is the business that consumes those services provided by/from the data centre.

Now the important figure they need from us in order to do their value calculation is cost...how much did it cost to deliver that service from the data centre? Not just the metered power cost (although that is an element of it), but the actual fully loaded cost with overheads to deliver.

The truth is we're going to need to become rather good at accounting for those costs, explaining them and justifying them because we're going to get asked, sooner or later.

I believe the productivity discussion is a simply a distraction that we as an industry have become fascinated with. It is just another mask to hide the true leveller that will define the winners from the losers. The ability to cost account. It's hard, there is no doubt, but it is possible and it won't be long before being able to per service cost account will re-define the leaders from the rest of the pack.