Energy costs in network can be greater than energy savings in the cloud

[Excellent paper by J. Baliga and Rod Tucker that shows focusing on energy consumption of “green” data centers may be misleading. Data transfer over a routed network could be significant component of the overall power consumption. Optical networks of course a much smaller energy footprint than routed networks. Of course if the routing and computing nodes are powered by renewable energy then the CO2 footprint is very small (although overall energy consumption may be in fact higher) –BSA]

http://www.hpcinthecloud.com/news/Lost-in-Transport-Why-Cloud-Isnt-Always-the-Greenest-Option-104702039.html?utm_source=twitterfeed&utm_medium=twitter

Researchers from the University of Melbourne in Australia have come to the conclusion that cloud computing is not always the greenest option on the storage and processing as well as the software level. This research examined the issue in both the public and private cloud context in comparison to the energy consumption used for the same tasks on a local system.
The authors argued that most studies seeking an answer to a similar question about the “green” nature of the cloud have only looked at the datacenter’s energy consumption and have thus failed to include the important issue of energy use during data transfer. They suggest that the transport of data to and from datacenters, particularly since public cloud center might be a continent away, uses quite a bit more energy overall than simply storing data locally.
PhysOrg.com reported that, “for cloud processing services (in which a server such as Amazon Elastic Compute Cloud processes large computational tasks only and smaller tasks are processed on the user’s computer) the researchers again found that the cloud alternative can use lower consumption only under certain conditions.” This is because “the large number of router hops required on the public Internet greatly increases the energy consumption in transport, and private cloud processing requires significantly fewer routers.”
The leader of the research project, Rod Tucker, told PhysOrg.com that when one is using the cloud for data storage (for instance on Amazon’s Simple Storage platform) cloud uses less energy than typical computing, but only when that service is used infrequently and not in a high-performance context since data transport energy use is minimal.
While the study focused on more garden variety processors and systems common for desktop users, this research might lend some insight to larger enterprise centers that are reliant on the cloud for some or all of their business operations. While many enterprise users might look at their bottom line before analyzing their overall carbon footprint, a study on the large enterprise scale that takes data transfer into account to offer a “green” score for a company might be a good idea.
Making the process of data transport more energy efficient needs to become a priority, but luckily there are incentives to do so. While the end user might not be bearing much of the cost of inefficient data transfer consumptions, it is in the best interest of cloud providers, who must remain competitive via pricing models, to constantly improve this critical aspect of their datacenters.
The research from the University of Melbourne will be published soon from Jayant Baliga and colleagues. The paper is called “Green Cloud Computing: Balancing Energy in Processing, Storage and Transport” and will be published in the journal Proceedings of the IEEE.

http://people.eng.unimelb.edu.au/rtucker/publications/files/Baliga_Ayre_Hinton_Tucker_JRLStrTrans.pdf


------
email: Bill.St.Arnaud@gmail.com
twitter: BillStArnaud
blog: http://billstarnaud.blogspot.com/
skype: Pocketpro