Counting the Cost Through RAI in Data Centers

By CIOReview | Thursday, March 27, 2014

FREMONT, CA: Usage based pricing model is one solution for optimizing data center operation. To realize that goal, a data center’s resource supply and demand sides should match as closely as possible and for this purpose a metric, namely resource allocation index (RAI), to measure the matching between demand and supply sides of data center resources, has been proposed by Rajat Ghosh of Georgia Tech.

For every computation happening in the data center, there is a combination of volume servers, network switches and storage disks involved in it. All these equipments require electricity to stay up and running. While all these equipments need not necessarily be used at all the times, there is no discount in the amount of power these equipments consume as all the devices are run regardless of whether there is any processing happening in the servers or event of storage access transpiring. This calls for a major investment of time and thought over the electricity (energy) consumption which also becomes the fundamental resource of the data centers.

The data center’s electricity should ideally be consumed by its IT equipment however surveys indicate that a large amount of data center’s electricity is utilized by its cooling hardware like server fans and cooling towers. There is a metric called Power Usage Effectiveness (PUE) to measure power utilization but it concentrates on the supply side of the value chain for Internet Data Center (IDC). To address this issue the introduced metric, RAI helps to regulate power consumption.

RAI estimates the amount of electricity essential for a data center to serve one request. RAI can also be used to evaluate the resource allocation performances of multiple data centers. Lower RAI value is advisable as with low resource data centers can satisfy its demand. However too low shows that the given data center is not drawing the requisite electricity to support its demand,  while a higher RAI value indicates over-provisioning that would lead to waste of electricity.

Therefore, a data center’s RAI value indicates whether its operating resources have been allocated in one of the following three modes: over-provisioned, under-provisioned, and optimally-provisioned.