Should the IT department be accountable for energy use?
TCO? What TCO!
Reg Tech Panel As the green debate rages on, one of the findings that came from our recent survey of Reg Readers revealed that despite all the talk about energy savings and IT, the majority of IT departments are not accountable for spending on electricity to run systems.
The survey also told us that a lot of IT departments would have a problem if they were made accountable, as their ability to measure and monitor energy consumption by IT is limited in most cases at the moment.
Practical limitations to one side for a minute, though, the bigger question is: Should IT be made accountable at all? Are we potentially opening a can of worms, or does the nature of accountability need some real hard thinking, as opposed to kneejerk reaction? The suggestion "Let’s make the IT department directly responsible for the electricity bill relating to IT" doesn’t even begin to describe the actual challenges and practical realities behind the sentiment, does it?
And what of those practical limitations? We’ve seen many times the effects of policies that are backed by little or no investment into actually allowing said policies to function, so while organisations kid themselves they have made a difference, the actual day-to-day improvements are minimal.
Total cost of ownership (TCO)
If we think about a plot of ‘influence v proximity’ then at least we could start considering the areas that are actually under ‘direct control’ (lighting, heating, cooling, running equipment inside the data centre), versus areas which are under ‘indirect control’ – because they simply need to be ‘on’ in order to allow delivery of whatever the data centre is pumping outwards (networks, substations, routers, switches), versus areas which are under ‘pseudo control’ (desktops, printers etc – network devices) which IT can see and change, but whose actual operation isn’t readily controlled without investment in more kit and policy/process training for users.
The idea that IT should be responsible (for example) for the cost of delivery (‘down the pipe’) would be like replicating the business models used by utilities companies before deregulation and vertical/horizontal integration realigned business models. These days our electricity/gas/phone bill factors in the cost of transmission, but the performance of the pipe is not the responsibility of the supplier; it’s the responsibility of the carrier and the owner of the local loop.
True, the services IT delivers would be utterly pointless without the network to deliver them. So rather than simply lumping everything into one pile, what about creating zones, which would have different levels of priority when it came to energy consumption? Somebody needs to take responsibility for them of course, but the task of management/compliance would become easier if the organisation had helped classify and define them, rather than IT simply being told to ‘spend less’. Otherwise we run the risk of simply setting IT up as the scapegoat for yet another ‘big idea’. The zoning idea would at least focus the organisation's mind on what might be practically achievable in assessing total cost of ownership (TCO). The area closest to the IT department – the data centre – would of course be first up; and it already is, so we’ve already come full circle.
Or is this all nonsense, and are we doing the IT department a disservice by ignoring the fact (as mentioned in responses to previous articles) that part of their job already is to find the right balance between performance and cost? Should we be asking harder questions of all hardware vendors, not just the server vendors, or should the IT department be spending their lunchtimes at the local electrical shops buying amp/voltmeters?
The question still remains – do we want our IT departments to actually spend their valuable time thinking, planning and implementing this stuff, and if so, what jobs are we happy for them to forego while they do it?
As usual, tell us what you think – we're running The Register Green Computing debate on the 27th February and the more ‘practical ammo’ we have lined up, the more interesting it will be.
Sponsored: Hyper-scale data management