Multi objective heuristic algorithms for scheduling workflow applications in heterogeneous cloud

Abstract

Workflow applications encompass a sequence of tasks which enables the analysis of data in a well-structured and distributed manner. These tasks are interdependent thus leading to a huge amount of data exchange between preceding and succeeding tasks during workflow execution. Many scientific research areas such as biological engineering, ocean sciences, earthquake science and many other promising areas include processing of workflow application having Terabytes or Petabytes of data. Processing and analyzing such data involves more complex computational resources which would be too intricate for the traditional computing environment. Considering the scalability feature of cloud computing environment that offers unlimited resources for execution, the aforementioned issue could be solved by implementing the workflow applications in the cloud. Cloud computing, universally known as cloud , is a popular Distributed system paradigm which offers utility-based IT services on-demand to the users on a pay-per-use basis. Thus, it facilitates the execution of workflow applications with low cost by avoiding the need of owning any infrastructure. In specific, Infrastructure as a Service (IaaS) provides a scalable and flexible cloud assets for the deployment of workflow applications by offering access to heterogeneous infinite resource pools. This study aims at developing multi-objective scheduling algorithms for mapping workflow applications on to heterogeneous resources. The thesis presents four different algorithms for scheduling which are capable of satisfying a set of Quality of Service (QoS) parameters expressed in terms of execution time, cost, energy, and resource utilization. newline newline newline