At the end of last year an influential article appeared in the New York Times entitled ‘Power, Pollution, and Internet Waste’, focussing on the wastage of energy in the Data Center Industry. Here, Soeren Schroeder, software director at Schneider Electric, analyses whether this interpretation of the industry is justified, and the steps we can take to ensure greater energy efficiency in Data Center Infrastructure Management (DCIM).
The full presentation can be viewed as a live blog-cast here.
Data Center Energy Consumption
One of the ‘facts’ that was brought forward by the New York Times Article was that about 90% of data center energy was really wasted somewhere in the process because of the way that we do business or the way that we are handling these data centers. I think it becomes a bigger problem and is more visible typically for high density data centers. But again, the problems remain the same no matter what size data center you are operating.
We are wasting too much energy in the industry today, New York Times article or not, and there are a lot of things contributing to the mix.
General Factors Contributing to Data Center Energy Waste
Redundancy is a big factor, and the degree of back-up we build into our data centers to compensate for it. We typically measure the quality of our data centers in terms of how many numbers of 9 in uptime that we can accomplish, and that is really in contrary to the wish and requirement of re using energy.
Legacy wise we also had an approach of putting one application on each server – a really wasteful way of doing things that thankfully we have moved away from. We also have to admit that we have had an industry where energy has been relatively cheap for us, so we didn’t have to bother about saving on the energy side of the equation.
Stranded capacity is one of the really big energy wastage issues you should try and fight in your data center. You fill up your rack with all the servers you can physically put in there, but you still have 2KW of power remaining that you don’t need for any of those servers. Well that capacity, it might be that you can utillisie it somewhere else but really in essence it has been allocated to that rack by definition. It becomes much worse when we start talking about cooling capacity because what type of capcity do we really have available and do we really have it available where it is?
So those are some of the dilemas we are talking about, and when you’re talking about cloud computing in all its different shapes and forms, we really have another shape of complexity. In all honestly I think we have a big dilemma in our industry.
Data Center Infrastructure Management
DCIM is typically presented as a software package or a software suite that allows us to take control of the business processes around our data centers all the way from deploying equipment into the entire workflow. DCIM is typically implemented as a software suite, but really from an organisational point of view it’s much more a statement, more of an understanding and an adoptive methodology for how you manage your data center.
From a Schneider point of view, we have an interesting legacy in this business because our DCIM StruxureWare for Data Centers is typically not a purpose built operational tool. It’s actually a long evolution of an internal design tool originally launched more than 15 years ago to internally help our own engineers in the planning, designing and building of data centers. Providing KPIs around the data center operation in terms of energy use and energy spend. Remember it’s not the software that makes the difference to you, it’s really your own attitude, your own processes, and how you go about that.