Studies on performance analysis of cloud computing based data centers with queuing models using MATLAB

Abstract

Cloud computing was the technology developed to store the data and support the users with the access to the data stored by charging a minimal amount for the storage of data and for providing necessary steps for storing the data and fro providing security to the data that was stored. The content stored in various servers at various locations based on the type and size of the content. The content can be accessed to the users with valid registrations and a set of security verifications entered by the customers. The content that was hosted in the servers can also be used for hosting various applications and various other set of options of systems in various fields. It is one of the most famous and mostly used research areas in the recent years for further development in various set of applications and its usages related to several set of customers in the real time environment. Performance evaluation in cloud computing has been another major thrust area in the recent past, which is of crucial interest for both cloud providers and cloud customers. Only few notable works have been published with regards to performance evaluation in cloud computing. Generally analytical models established for assessing the working and the performance of cloud server farms can be studied under variety of configurations and assumptions are based on queuing theory and its accuracy is verified with numerical calculations and simulations. The problems at hand give rise to the task of evaluating the performance of data center with various queuing models to understand the distribution of the performance parameters with arrival and service rates, traffic intensity, number of servers and the associated probabilities. The goals of this thesis is to provide a framework through programs related to queuing models and evaluate the performance parameters, attempt validation, sensitivity analysis and make comparisons for data centers. Present thesis evaluates the performance parameters of cloud data centers based on queuing theory for both single server and multi-server models. The steady state performance parameter formulations identified are programmed in MATLAB® environment. The models considered for evaluation for single servers include M/M/1, M/G/1, M/D/1. Service rates have a wider range of distributions including exponential, generalize and Erlang type.