A performance analysis of a typical server running on a cloud

Abstract

In recent years, Cloud computing has become very popular from individuals to big enterprises because of the reduced cost, minimal management effort etc. Cloud providers like Amazon, Google are now offering resources for web deployment, storage, servers etc. However, due to the varying load and cost, evaluating the performance of task scheduling policies in these real Cloud environments is very challenging. In this paper, the performance of Earliest Deadline First (EDF) scheduling policy has been investigated using CloudSim and the hardware configuration of Amazon Web Services (AWS) and Google Cloud Platform (GCP) with the time-out of a web and FTP servers. In addition, a comparison between space-shared and time-shared task provisioning policies have been examined which shows that the average execution time can be minimised by using space-shared policy in both AWS and GCP.