Performance Modelling Power Consumption and Carbon Emissions for Server Virtualization of Service Oriented Architectures (SOAs)

摘要

Server Virtualization is driven by the goal of reducing the total number of physical servers in an organisation by consolidating multiple applications on shared servers. Expected benefits include more efficient server utilisation, and a decrease in green house gas emissions. However, Service Oriented Architectures combined with Server Virtualization may significantly increase risks such as saturation and Service Level Agreement (SLA) violations. Since 2006 National ICT Australia Ltd (NICTA) has been developing a technology for the performance modelling of large scale heterogeneous Service Oriented Architectures (SOAs). The technology has been empirically trialled, refined and validated with collaborating Australian Government agencies to address critical performance risks. Many government SOAs are developed, tested and deployed on virtualized hardware, and we have developed the capability to model the performance of SOAs deployed on virtual servers. In this paper we provide an overview of NICTA's SOA performance modelling approach, and then explore a number of alternative deployment scenarios for an example SOA based on a synthetic carbon emission trading system. We show how our modelling approach provides insights into the relationship between workloads, services, and resource requirements, and can therefore be used to predict server power consumption and carbon emissions. We model and evaluate four different deployment options including planned and optimised resources, server virtualization, and computing-on-demand (cloud computing using Amazon EC2). We conclude with an overview of other potential problems and benefits of SOA virtualization.