Regardless of platform, one of the big problems that cloud engineers encounter is the under-performing of virtual machines. This phenomenon can occur when running a public, hybrid or private cloud. When nodes under-perform, it wastes both precious computing power and the energy that it takes to run that node. The difference may seem insignificant when you look at it on an individual virtual machine basis. However, think about an entire datacenter that houses a couple million virtual machines. In this case, a sizable percentage of under-performing virtual machines equates to an organization that is spending money that they shouldn’t have to spend on the cloud. Computer scientists at Stanford have developed a new method of routing cloud computing loads to servers that are best equipped for the action.
Led by Christos Kozyrakis, the team of computer science researchers found that application developers often overestimate the resources that they need when they develop an application. As applications change and receive updates, they can sometimes use fewer resources than originally anticipated which idles virtual machines ever more. This often results in virtual servers that have an average resource utilization rate of about 20% per server. The answer to this problem is named Quasar. Christos Kozyrakis and his research team feel as if Quasar will help datacenters utilize their resources by using an algorithm that helps match cloud computing loads with virtual machines that are running more efficiently.
Analysts have compared this same approach to the algorithms that are used by Google for search engine results. In a quote provided by the NY times, Kozyrakis mentioned that “They don’t wait for you to load 100 of your choices. They learn from what other people did. We computer scientists like to think we’re unique, that our programs are unique. But really, we’re not.” Quasar has been in development for over 2 years, and news outlets mention that the group has been approached by several commercial ventures that may be interested in acquiring the Quasar technology for their cloud loads. Quasar highlights a very important trend in the cloud. When everyone is using the cloud, the next best way to save money on your IT costs is to find ways that make your cloud applications run more efficiently.