Due to increasing power consumption and cost, data center energy conservation has become increasingly important. As a result, a number of algorithms for optimizing compute, storage, and networking have been devised. However, these algorithms focus on job distribution based upon optimizing energy efficiency, and do not take availability into consideration. For example, if there are 3 servers, allocating workloads based only upon energy efficiency may result in consolidation of the workload into only 1 server, resulting in a single point of failure. This paper outlines the importance of availability and designs a hill climbing algorithm to prevent failure zone failure and hence also reduce the energy consumption by the datacenter.

No abstract

No abstract