FOCUS UPDATE POWER
Issue 6, Oct/Nov 2009
Fixing the price of energy is favoured as a
short-term solution to data center power costs Fixed-price energy would be welcomed by the vast majority of data center operators, but there are conflicting views on optimum lengths of time for contracts. In the US, more radical solutions are being considered
information officers and director-level IT professionals’ preferences favoured shorter, fixed-term contracts of just six months.
“The popularity of price fixing is without doubt a reaction to the current economic climate,” adds Ruhan. “Yet this approach still fails to address the root causes of high power consumption within the data center. The data center industry needs to treat energy as though it were a currency and seek to get the best value from it.”
Over the next 12 months, providers will be looking ahead and coming to market with alternative energy sources for new data centers that will take them forward for the next three years.
In the meantime, however, the data center industry remains ill-equipped to answer the issues faced by end users and must raise its game in its level of understanding and ability to advise customers.
This is the world according to Andy Ruhan, managing director of Sentrum. There are clearly a number of factors, such as cost, legislation and carbon credits, he says.
“The problem with the data center industry is that standard data centers are by their very nature inefficient, but it is difficult to introduce something new into a tried and tested formula because you can’t point to empirical data on how effective any changes will be,” he says.
Sentrum decided to look into the matter further and commissioned research in the UK. It found in its report, Energy: The currency of the data centre, that while most IT professionals (99 per cent) would like to fix the price of the power supplied to the data center, the average period for wanting to do so was just 2.1 years.
Although some companies (16 per cent) would welcome such agreements across five years or more, Sentrum found that a significant proportion (46 per cent) would only welcome a tie-in period of less than two years. Chief
52
www.datacenterdynamics.com
One in two of those questioned in the report would welcome the ability to reduce the carbon footprint of their data center. The top topic for 60 per cent of the respondents was the requirement for improved automated reporting capabilities about their levels of energy consumption, mainly from existing hardware and software.
With this in mind, more than half of respondents (52 per cent) said they would welcome advice and recommendations from specialist consultants about energy consumption.
The research, however, also suggests that such recommendations are not often forthcoming from the industry, or can be inaccurate when offered.
“It is now down to industry players to work with customers to educate them on how to make the data centers more energy efficient and, crucially, cheaper to run,” says Ruhan. The research questioned more than 100 C-level UK executives.
DYNAMIC IT LOAD DISTRIBUTION IN THE US CAN SAVE MILLIONS IN DATA CENTER ENERGY COSTS
Research suggests that distributed systems should include electricity price fluctuations as a factor.
Today’s large companies that operate IT systems distributed over wide geographical areas have developed infrastructure that enables cloud computing. A study, published by a Massachusetts Institute of Technology
(MIT) graduate student suggests that the same infrastructure can and should be leveraged to take advantage of geographic and temporal fluctuations in energy prices.
Asfandyar Qureshi, who is working on a doctorate at MIT, and a team of associates analysed 39 months of hourly electricity prices from 29 locations across the US. To estimate the amount of savings organisations can expect from implementation of such dynamic IT load distribution, the team looked at 24 days of request traffic data for each Akamai server cluster in a commercial data center in the country. Akamai – a Cambridge, Massachusetts-based internet content delivery and application provider – provided request traffic data for the study.
Two observations make up the foundations of Qureshi’s argument: energy prices in the US vary dramatically on an hourly basis; and existing distributed systems already deploy dynamic request routing for purposes of fault tolerance and good performance.
In addition, many of the existing distributed systems are already capable of replicating data for processing requests at multiple locations.
The study suggests that organisations can save between two and 45 per cent of their energy costs by constantly moving IT load to different locations according to price fluctuations. The amount of savings depends on bandwidth costs, client performance levels, energy elasticity and room for greater client-server distances.
Qureshi shows that at least a two per cent reduction in energy costs can be achieved by implementing his proposed strategy without associated increases in bandwidth cost or reduction in climate performance.
“Relaxed bandwidth constraints,” in combination with energy elasticity, can allow for more than 30 per cent savings on energy cost. The study defines energy elasticity as “the degree to which the energy consumed by a cluster depends on the load placed upon it”.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72