Power availability is a big problem effecting IT managers today and in this article, Phil Lydford, CEO at e-shelter, discusses the true extent of the problem, how companies should be planning for the future and how, with London 2012 fast approaching, spare capacity has been earmarked for the Olympics

If you are planning to build a new data centre in the next five years and haven’t reserved the power for it with your supply company – then you may have to think again.

The UK notoriously runs very close to maximum  power capacity on a day-to-day basis, but with demand for electricity from businesses rising (despite the downturn) and the 2012 Olympics only three years away, spare capacity for power-hungry data centres is in even shorter supply.  Many IT managers have only recently woken up to the fact that unless they have already allocated power with their supply company for data centre developments the chance of getting it is slim.

You start to appreciate the scale of the challenge faced not only by IT professionals in the UK, but across the whole continent, when you consider that for 2007 the estimated power consumption for Western Europe was 56TWh (Terawatt-hour) and it is expected to rise to 104TWh by 2020.

Distribution problems
Even if you have reserved capacity for a planned development within the next few years, there is still a chance that the power infrastructure feeding your site may not be up to the job.  And, if that is the case, you could still find yourself without power for at least 18 and 24 months.  The problem is that the UK’s power distribution network has suffered from a severe lack of investment since deregulation and that means that many parts of it are not capable of transporting the large amount of power that is needed for a modern IT facility such as a data centre. 

Additional sources?
Renewable energy has been mooted by the IT press as a possible source of additional power capacity for the last couple of years but, in reality, the volumes required by even the most efficient data centre make solar, wind and even tidal energy wholly unsuitable at present.  You could look to refit an existing data centre site to increase, but even using virtualisation, the business will probably still require additional storage capacity over the next few years; plus the risks associated with an in-service upgrade are, arguably, far too great.

Another option is to improve the efficiency of an existing data centre.  It needs to be the job of everyone involved in the construction and operation of data centres to work together to maximise the efficiency of their facilities.  Using the latest systems and processes it is possible to make considerable efficiencies in an IT facility. 

More MPG
The first step is to calculate the PUE (Power Usage Effectiveness) of the facility.  Historically, data centres have been designed with large tolerances for operational and capacity changes, using practices that are now outdated. This leads to power consumption inefficiencies, with only a fraction of the grid power consumed by the data centre actually use for powering IT systems.  PUE is a metric that provides IT managers with an indication of how much power is being used in their data centre in relation to its equipment load and how much is being used by ancillary services.  It is much like that of miles per gallon (MPG) calculations for a car – the owner of the vehicle needs to know how efficient the engine is working so that they know how much petrol it is using.  A data centre manager can use PUE in the same way to assess the performance of their site(s).
The PUE of an old “legacy” can be as high as 5.0.  In comparison, new purpose-built sites that can be as low as 1.2.  In terms of power availability, a low PUE value means two things; less power is needed to operate the site, or less of the available power is being used to power non-IT equipment.  A decision can then be made to reduce power consumption (reducing the infrastructure needed to support it) or to utilise more of the available electricity for core IT functions – effectively, increasing an organisation’s potential processing power.

The final option is to use specialist data centre providers to help source additional space. IT managers are under a lot of pressure from the board to get more for less and this route may prove more cost and time efficient than researching, negotiating and building a brand new facility.  There is also the question of finance - getting funding for new capital projects is tough in the current climate – and senior executives will look kindly on any projects that can reduce capital spend in favour of operational expenditure right now. 

With regards to power availability, it is likely that any announced new build will have already reserved enough capacity to at least support the anticipated capacity of its facility – with a built in contingency for supporting additional capacity.  In addition, forward thinking data centre space providers will also have custom designed their facilities to the very latest specifications - ensuring that it makes best use of available power and that dependent upon the hardware its clients put into it, the data centre will always have the lowest possible PUE ratio.

Regardless of what route you decide to take, a new data centre needs to be a strategic decision, not one made to solve a short term problem. Considerations like power usage effectiveness, location, communications, power availability and funding need to be carefully thought out if a facility is to provide the computing capacity to support the business and deliver best value.  Failure to do so could lead to an inefficient facility that isn’t fit for purpose or, worse still, investing in the building of a brand new data centre, only to find that there isn’t enough electricity available to power it.