The operating cost of modern data-centers is the primary concern when considering “cloud” based algorithms. These data-centers contain numerous machines that run distributed, graph based algorithms in order to fully take advantage of the available resources. We attempt to lower the operating cost of such systems by leveraging the graph structure during the testing phases of of a software release cycle. Applying this technique involves modifying typical system hypervisors and the ability to support a multi-staged environment. We use the term multi-staged to concisely describe any composite environment consisting of a development environment, a QA environment, a staging environment and the production environment. We verified this result by running a sample application on both architectures and measuring operating cost of the underlying systems. Our tests confirm, for this case, that operating costs can be reduced significantly by this modification of data-center hypervisors.
Nate Woods
