HySARC2: Hybrid scheduling algorithm based on resource clustering in Cloud environments

Abstract

Cloud Computing is a fairly new paradigm but evolving very fast. Nowadays, business enterprise services and middle ware such as SAP or Oracle are integrated into a Cloud infrastructure. Also, Cloud services are used for computational or IO intensive applications from multiple science fields like physics, microbiology or weather forecast, due to the large amount of resources available. In this context, it must be paid attention to resource utilization in Cloud environments. Therefore, the scheduling of tasks on Cloud resources is a core issue, with impact for users and service providers. We proposed HySARC 2, a novel scheduling algorithm based on traditional approaches, which considers clustering of the available resources in the infrastructure in the phase of resource allocation. The resources clustering into groups is used by our proposed algorithm in a hierarchical way, executed in two phases. First, tasks are assigned to groups of resources and further, in a second phase, inside each group of resources a classical scheduling algorithm is executed. The proposed algorithm is suitable for heterogeneous systems and sets of applications with various requirements (both IO and computational intensive). © Springer International Publishing Switzerland 2013.