Estimating the processing time of a model of cloud computing

Abstract

In recent years, cloud services have been used not only as storage but also to increase computational power. Cloud computing systems perform computations using distributed computation across a large number of computers connected to the Internet. Therefore, cloud computing systems are designed to compute correct results even if failures occur on some of the computers. Many cloud computing systems have mechanisms for redundant calculation for this purpose, which makes it difficult to estimate the computing power of cloud services. In this paper, we propose a method that can estimate the computing power of cloud computing systems. First, we define a simple model of cloud computation. Assumptions about hardware and software control of the model are shown. We introduce a concept called the round number. We show that the round number is crucial for the overall computation time, and derive a formula for the distribution function of the round number. We show that the overall computation time of cloud computing systems can be approximated using the distribution function of the round number combined with an approximation of the frequency of each particular round number’s computation time. Simulation of distributed computation is performed for an experimental problem, and the validity of the proposed method is confirmed.