Performance Evaluation of Multi-Core Multi-Cluster Architecture (MCMCA)

Introduction

Chen, Wills, Gilbert, & Bacigalupo (2010) define cloud computing as an emerging business model that delivers computing services over the internet in elastic self-serviced, self-managed and cost-effective manner. Cloud computing doesn’t yet have a standard definition, but a good working description of it is to say that clouds, or clusters of distributed computers, provide on-demand resources and services over a network, usually the Internet, with the scale and reliability of a data centre. Cloud computing provides a pool of computing resources which includes network, server, storage, application, service and so on that is required without huge investment on its purchase, implementation and maintenance that can be accessed through the Internet. The basic principle of cloud computing is to shift the computing tasks from the local computer into the network (Sadashiv & Kumar, 2011). Resources are requested on-demand without any prior reservation and hence eliminate overprovisioning and improve resource utilization.

Cloud computing has changed the way both software and hardware are purchased and used. An increasing number of applications are becoming web-based since these are available from anywhere and from any device. Such applications are using the infrastructures of large-scale data centres and can be provisioned efficiently. Hardware, on the other side, representing basic computing resources, can also be delivered to match the specific demands without the user/consumer having to actually own them. As more organisations adopt cloud, the need for high availability platform and infrastructures, the cluster, to facilitate and distribute the load across multiple processors is evolving (Chang, Walters, & Wills, 2014). The deployment of clustered applications in Cloud infrastructures supports the capabilities of resource configuration and ensures communication between shared resources (Kosinska, Kosinski, & Zielinski, 2010).

The emergence of High Performance computing (HPC) that includes Cloud computing and Cluster computing has improved the availability of powerful computers and high speed network technologies. It can be concluded that the main target of HPC is better performance in computing. HPC aims to leverage cluster computing to solve advanced computation problems. While cluster computing has been widely used for scientific tasks, cloud computing was set out for serving business applications. Dillon et al. (2010) have pointed out that the current cloud is not geared for HPC for several reasons. Firstly, it has not yet matured enough for HPC; secondly, unlike cluster computing, cloud infrastructure only focuses on enhancing the overall system performance; and thirdly, HPC aims to enhance the performance of a specific scientific application using resources across multiple organisations. The key difference is in elasticity, where for cluster computing the capacity is often fixed, therefore running an HPC application can often require considerable human interaction, such as tuning based on a particular cluster with a fixed number of homogenous computing nodes (Schubert, Jeffery, & Neidecker-Lutz, 2010). This is contrasted with the self-service nature of cloud computing, in which it is hard to know how many physical processors are needed. In order to achieve higher availability and scalability of an application executed within cloud resources, it is important to supplement the capabilities of management services with high performance cluster computing to enable full control over communication resources.