DWCSO: A Hybrid Intelligent Algorithm for Multi-Type Workload Scheduling in Cloud Computing

Abstract

Cloud computing is the emerging technology used by almost every industry in the era of artificial intelligence. Small to large organizations are continuously moving towards cloud-based systems for ensuring easily availability of data, fast computation and other pay-per usage IT sector services in remote areas. As a result the workload of cloud data-centers is increasing regularly as the cloud technology provides fast computational pay-per-use services. So, resource management and scheduling have been vital issues for research since last decade. The data-centers need to be maintained at various stages like network, cooling systems, servers’ maintenance etc. One of the major mechanisms of resource management is task scheduling. Various task scheduling techniques using swarm intelligence have been developed so far. However, these techniques suffer from either exploration or exploitation. So, this paper proposes a hybrid DWCSO tasks scheduling algorithm which is based on Differential Evolution (DE), Adaptive Inertia Weight (SMIW) and Cat Swarm Optimization algorithm. The SMIW inertia weight balances the seeking and tracing modes. The DE based operators help the DWCSO algorithm to escape from the local stagnation. The proposed algorithm has been tested on various scientific workflows and Google Traces 2019 independent-tasks datasets. After evaluation, it is observed that the proposed DWCSO algorithm consistently outperforms its competitor IGWO, achieving 9.41 to 17.33% improvement in makespan, 4.93 to 12.56% in cost with higher throughput in case of workflows. In case of Google Traces 2019, the proposed DWCSO achieves 15.52 to 19.32% improvement in makespan, 8.05 to 10.20% in cost over IGWO with higher throughput. The maximum improvement observed against PSO is demonstrating its superior efficiency.