• Data center news, conferences and training
• Conferences
• Training
• Magazine
• Video
• Census
• Awards
• About Us
• MyDcD

If you look at the data-centre TCO landscape in the UK the cost of electrical power now exceeds 50% of the total costs over 10 years.  With UK power now around the GB£0.09/kWh range it will be similar for the major data-countries in Europe although the US will be someway behind us on the cost curve.  In the UK we have power cost inflation predictions of 15% per year (or 80% by 2016, you can take your pick) that will only serve to increase that 50% proportion as the cost of land, construction, IT hardware and operations is not set to grow much above the background inflation rate.  That said, I was pondering on the problem of latency.  Some trading banks and other organisations need to have their electronic trading platforms within a couple of milliseconds of the markets.  For the UK and the London Stock Exchange that scribes a circle of possible locations (based on the speed of light, the speed of the gateways and the software/hardware) with most peoples estimations being 10-15km as the maximum.  That ‘forces’ the user to build their trading platform (and the data-centre that houses it) slap-bang in the middle of the prime real estate of SE England, if you are lucky just outside the M25 car-park and if you are luckier somewhere that you can actually source the power you need.  But increasingly I hear talk that, under cost pressure, systems are being planned that put just the sharp-end of the platform (the essential 150-250kW) close to the market with the back-office being freed-up to relocate to pastures cheaper.  That might solve two power problems (cost and availability) but what about the ‘server hugger’ syndrome, will that ever decline?  We used to say that two main factors decided a good location for a data-centre; power & connectivity.  To those, should a third factor always been added? Client-proximity.