LiiT
Laboratory for Internet and Innovative Technologies
Abstract

Cloud computing offers scalable hardware resources on Infrastructure-as-a-Service (IaaS) service layer. However, instances of virtual machines (VMs) are often manually initiated by cloud clients producing degradation of service availability and service latency in peak loads. In this paper we propose a solution that handles the compute peak loads dynamically for web services hosted in cloud. Our solution introduces a middleware layer between clients and server. The middleware layer will instantiate additional VMs dynamically on demand as service load reaches defined minimum performance level and will forward the messages across VMs. The additional VMs will be shut down when service load returns to defined nominal value.

Authors

Sasko Ristov, Marjan Gusev, and Goran Velkoski

Keywords

Cloud Computing, Performance, Virtualization, Middleware

Full Paper

The technical report is published as a paper in Proceedings of the 9th Conf on Informatics and Information Technologies, (Editors Bakeva and Gjorgjevikj) CiiT 2012, Bitola, Macedonia, 2012