Stanford Researchers Increase Cost Effectiveness of Cloud and Contribute to Cleantech

Add Your Comments

Companies are always looking for ways to reduce costs. Researchers at Stanford may have developed a way data centers and cloud providers can save money using existing infrastructure.

Christina Delimitrou and Christos Kozyrakis looked at server utilization at Twitter for a month. They found CPU utilization was below 20 percent as needed resources were being overestimated.

In a Stanford Daily article, researcher Delimitrou said, “If you measure the utilization [of the current system], it’s at around 10 to 20 percent, which means that about 80 percent of the computing power that you could be getting out of the machines is being wasted.”

Their solution is an automated cluster management system called Quasar. Amazon and Netflix have similar algorithms to recommend items or movies to their users, but they are not using those algorithms to deal with system resource demand.

Rather than relying on estimation of how many resources are needed by users, Quasar instead asks how fast the needed applications should run. The software then translates speed into required resources.

“It can help the operators of cloud computing facilities serve more customers and more applications,” Kozyrakis said. “Quasar increases the cost-effectiveness of these facilities by two to three times.”

Companies also look for ways to differentiate themselves with consumers. Being perceived as a “green” company is one great way to do this. In 2012, The New York Times examined data centers for a year and found that of electricity pulled from the grid, data centers waste 90 percent or more of that power.

Future work by the Stanford team will continue to promote cleantech through increasing efficiency. Delimitrou told the WHIR “One aspect of future work revolves around the notion of hybrid clouds, that is systems where both privately-owned machines and machines on a public cloud provider are used. We are currently looking at how to provision each system and how to partition the load in a way that guarantees that certain performance constraints are met while being cost efficient for the end users. ”

Green data centers tend to focus on using green power and cooling. Since data centers consume an enormous amount of power each day, reducing this load would be environmentally friendly.

Another way to look at this is that fully utilizing the existing resources in an existing data center could reduce overall power consumption by mitigating the need to add more servers and build more data centers. Implementing Quasar could be a way for cloud and hosting companies to increase efficiency, reduce costs and distinguish themselves as a “green hosting” provider.

Add Your Comments

  • (will not be published)