An intelligent dynamic time efficient throttled load balancing algorithm for cloud computing environments

Abstract

Load balancing in cloud environments guarantees favourable response time, execution time, efficient resource utilisation and balanced  traffic and workloads for the servers. Cloud computing service providers are prone to lose clients when clients experience high response  time or delayed responses on their requests. This is due to inefficient use of artificial intelligence contributing to suboptimal load balancing algorithms in cloud computing environments leading to poor service delivery to cloud users. This research aims at building an  intelligent load balancing algorithm as an improvement of throttled load balancing algorithm. The proposed algorithm considers servers’  availability, time remaining to complete task processing in servers, and processing requirements for received tasks to determine optimum workload distribution for servers, resource utilisation, execution time and response time. The proposed load balancer receives  requests, determine the best server to process the task depending on service’s load requirements, server’s availability and processing  properties. If the determined server is available, it receives the service, else, the service is queued with a tag of the server to service the  task. If the server that is not the best at servicing the task is available, the task will be queued only if the time to wait for the best service  plus processing time on best server is less than the time it takes to process the task on the server that is available but not the best for the  task. Java programming language was used to implement the proposed algorithm in a cloud simulation platform. This research makes  use of CloudAnalyst which is a CloudSim extension, to conduct experimental comparisons between the proposed algorithm and existing  load balancing algorithms. This research observed that balancing the load in a Throttled load balancing algorithm improves resource  utilization at the expense of response time and execution time. The proposed algorithm outperforms throttled, round robin and equally   distributed load balancing algorithms in terms of response time, and execution time.