The Massachusetts Institute of Technology published an article discussing "energy-aware internet routing", which it argues could save data-intensive businesses millions of dollars every year in electricity costs. But as with most IT developments, the potential gains are not as easy to come by as headline statistics would suggest, with a number of issues that will limit the benefits available.
Working with distributed computing company Akami, the research suggests that business could reduce their data center power consumption by as much as 40% by re-routing tasks to locations where electricity prices are lowest on a particular day. By shifting the most computing-intensive processes to low-cost facilities, quality of service is maintained while utility costs are cut, creating something of a win-win for enterprises.
MIT noted that data center power consumption is likely to increase in the coming years, as more and more information and applications are moved into the cloud. For businesses with a heavy reliance on technology, the increased flexibility offered by distributed computing and web applications will be accompanied by higher utility bills, which is an issue that may not have so-far been considered.
Researchers analysed 39 months of electricity price data for 29 US markets, which found costs fluctuating due to seasonal changes in supply, fuel price increases, and changes in consumer demand. This led to "a surprising amount of volatility, even among geographically close locations" - no one place was always the cheapest, with large fluctuations exhibited on a short time-scale.
The MIT team devised a routing scheme designed to take advantage of daily and hourly fluctuations in electricity, also taking into account the costs incurred routing data to other sites when compared with the potential energy savings. Information collected from nine Akami servers across 24 days of activity provided a way to test the routing scheme using real-world data, which delivered results deemed "pretty surprising".
With an increased focus on corporate' green credentials, it is worth noting that the MIT proposal cuts energy costs, but not consumption. Savings could also be achieved by increasing data center efficiency through the use of technologies such as virtualisation, reducing total power use instead of reducing the price of the power consumed. Website TreeHugger suggested a variant of the MIT model could be developed to take into account green credentials alongside costs.
Other issues also have to be taken into consideration to accurately measure the benefits. The concept relies on servers using substantially less power when idle than when working at full capacity, which is not necessarily the case and is difficult to monitor across multiple sites, while it is also essential to have access to up-to-date pricing information. In addition, capacity needs to be available at the data centre with the lowest-priced electricity, and there is still the issue of large amounts of data either needing to be replicated or moved between sites to reap the benefits.
Working with distributed computing company Akami, the research suggests that business could reduce their data center power consumption by as much as 40% by re-routing tasks to locations where electricity prices are lowest on a particular day. By shifting the most computing-intensive processes to low-cost facilities, quality of service is maintained while utility costs are cut, creating something of a win-win for enterprises.
MIT noted that data center power consumption is likely to increase in the coming years, as more and more information and applications are moved into the cloud. For businesses with a heavy reliance on technology, the increased flexibility offered by distributed computing and web applications will be accompanied by higher utility bills, which is an issue that may not have so-far been considered.
Researchers analysed 39 months of electricity price data for 29 US markets, which found costs fluctuating due to seasonal changes in supply, fuel price increases, and changes in consumer demand. This led to "a surprising amount of volatility, even among geographically close locations" - no one place was always the cheapest, with large fluctuations exhibited on a short time-scale.
The MIT team devised a routing scheme designed to take advantage of daily and hourly fluctuations in electricity, also taking into account the costs incurred routing data to other sites when compared with the potential energy savings. Information collected from nine Akami servers across 24 days of activity provided a way to test the routing scheme using real-world data, which delivered results deemed "pretty surprising".
With an increased focus on corporate' green credentials, it is worth noting that the MIT proposal cuts energy costs, but not consumption. Savings could also be achieved by increasing data center efficiency through the use of technologies such as virtualisation, reducing total power use instead of reducing the price of the power consumed. Website TreeHugger suggested a variant of the MIT model could be developed to take into account green credentials alongside costs.
Other issues also have to be taken into consideration to accurately measure the benefits. The concept relies on servers using substantially less power when idle than when working at full capacity, which is not necessarily the case and is difficult to monitor across multiple sites, while it is also essential to have access to up-to-date pricing information. In addition, capacity needs to be available at the data centre with the lowest-priced electricity, and there is still the issue of large amounts of data either needing to be replicated or moved between sites to reap the benefits.
-