Characterizing Failure and Repair Time of Servers in a Hyper-scale Data Center

Abstract

Hyper-scale data centers are used to host cloud computing interfaces to support the increasing demand for storage and computational resources. For achieving specific service level agreements (SLA), this infrastructure demands highly available cloud computing systems. It is necessary to analyze the server failure incidents to determine the way of improving the reliability of the system since the computational interruption causes financial losses for the data center owners. Regarding the reliability analysis, it is important to characterize the time to failure and time to repair of the servers. In this paper, a publicly available data set from Google cloud-cluster data center will be analyzed to find the distribution function for the time to failure and the time to repair for the servers in a cloud based data centers.