Arrival based deadline aware job scheduling algorithm in cloud

Abstract

This study designed a scheduling algorithm for Cloud Computing Environment by taking values of different parameters from a file provided by a user which gives the details of different cloudlets to be made. The cloudlets are assigned to different hosts according to the Shortest Deadline First combined with First Come First Serve where the Cloudlets are being submitted at different arrival times. The deadline provided by user is used to determine whether the Cloudlet is able to finish execution in the prescribed time interval or not. This implementation increased the number of jobs that finished executing in a particular time interval, thereby minimising the number of cloudlets that missed the deadline as compared to the Datacenter Broker scheduler which implements First Come First Serve. The characteristics remained fixed whether its for DataCenter Broker Scheduler or the Arrival based Deadline First Scheduler except the number of Cloudlets were increased gradually and the results were compared. The results show that the First Come First Serve Scheduler misses more Cloudlets from deadline as compared to ADSF Scheduler. All the cloudlets arrived at different time and were scheduled and submitted according to that time and not simultaneously which resulted in more efficient results. The results show that the Total De-layed Cloudlets have a huge difference for both the algorithms. Although, the waiting time is some what same.