Critical Analysis of Different Min-Min Task Scheduling Approach for Cloud Environment

Abstract

Cloud computing is a revolution in information technology industry because of its performance, accessibility, low cost and many other services compared to traditional online computing. Users are offered different level of services based on the quality levels mentioned in service level agreement. So cost of service also depends upon the type of resources provided to user so that there should be no violation of service level agreement. To provide the services as per service level agreement, there is need of efficient load balancing approach. The basic Min-Min algorithm was very easy and gives better plan and keeps the completion time of task very small relative to other algorithms found in the literature. But the major limitation of Min-Min algorithm is that it causes unbalancing of system. In the present work, authors have proposed a deadline and priority aware Min-Min load balancing algorithm which not only consider the deadlines but also consider the tasks priority. It has been analyzed and tested critically using MATLAB simulator on the various relevant parameters. Through simulation results, it has been found that deadline and priority aware Min-Min approach outperforms the basic Min-Min approach on all the relevant parameters.Keywords: Task Scheduling, Min-Min, Priority, Deadline, Load Balancing, Cloud Computing, Matlab I. INTRODUCTIONCloud computing is a new way of providing computing resources and services in real time environment. Basis of this technology is virtualization, elasticity and scalability. Cloud model provides on demand services to the user. User demands for different services at different time from different regions. Cloud is subject to user requirements, load balance, and other constraints that have direct effect on user consumption of resources controlled by cloud provider. So it is very difficult to conclude how many resources will be sufficient to provide the services as per service level agreement [1,2]. It becomes increasingly difficult to predict the demand requirements and load coming to resources. Load scheduling algorithm is a way to decide that which tasks are mapped to which resource. It is a way through which it is decided that which tasks will be mapped with which resource [3]. Load balancing is a hard problem as it involves the multiple tasks and resource combinations and among that best combination satisfying different constraints is selected. The prime objective of load balancing approach in cloud is to minimize the task completion time, task waiting time, makespan, optimize the utilization of resources and provide the quality of services as per service level agreement. It is also desired that load balancing approach will handles the higher priority tasks more efficiently relative to lower priority task. Unfortunately different techniques found in literature fail to satisfy all these objectives collectively [4,5]. In this paper, authors improvise the basic Min-Min algorithm by using priority-aware and deadlines for critical tasks. Proposed approach has been simulate...