Energy efficient and latency optimized media resource allocation

Abstract

Purpose This paper aims to provide a method for media resource allocation in Cloud systems for supporting green computing policies, as well as attempting to improve the overall performance of system by optimizing the communication latencies. Design/methodology/approach A common method for resource allocation is using resource agent that takes the budgets/prices of applicants/resources and creates a probability matrix of allocation according to the policies of system. Two general policies for optimization are latency optimization and green computing. Presented heuristic for latencies is so that the average latencies of communication between applicant and resource are measured, and they will affect the next decision. For gaining green computing, it is attempted to consolidate the allocated resources on smaller number of physical machines. So calculation formula of the price of each resource is modified to decrease the probability of allocating the resources on the machine with least allocated resources. Findings Results of proposed method indicates its success in both green computing and improving the performance. Experiments show decreasing 21.4 per cent of response time simultaneously with increasing tasks in the tested range. The maximum and minimum of saved energy is acceptable and reported as 79.2 and 16.8 per cent. Research limitations/implications Like other centralized solutions, the proposed method suffers from the limitations of centralized resource agent, like bottle neck. But the implementation of distributed resource agent is postponed to future work. Originality/value Proposed method presents heuristics for improving the performance and gaining green computing. The key feature is formulating all the details and considering pitch variables for controlling the policies of system.