Parallel Job Processing using Collaborative Time-Cost Scheduling for Amazon EC2

Objective: Cloud Computing is based on the pay per usage model. Amazon EC2 is the public cloud which provides IaaS using this model. Amazon EC2 provides virtual machines to the users. Cost for the use of virtual machines is based on the time for which it is being used. Amazon EC2 charges for partial instance hours even if the instances are idle. To reduce the cost of usage for customers, number of instances and the execution time must be reduced. Methods: In this paper we proposed a collaborative time-cost scheduling for parallel job processing. Our method aims to reduce the number of running instances to reduce the cost. As time is proportional to cost, jobs are processed in parallel. We designed a collaborative time-cost scheduling algorithm that selects the most suitable machine to run the job. Application: We developed a cloud data storage portal that enables users to upload, download, delete and compress large chunks of data on the fly without the need to download it to a local system and compress it offline. Findings: The status of the scheduling job is available to the user in addition to the status of the machine. Our algorithm uses minimum number of instances with no place for instance being idle. The time is reduced due to parallel job processing and cost is also reduced compared to sequential scheduling..