Home // ICAS 2014, The Tenth International Conference on Autonomic and Autonomous Systems // View article

Using Performance Modelling for Autonomic Resource Allocation Strategies Analysis

Authors:
Mehdi Sliem
Nabila Salmi
Malika Ioualalen

Keywords: Autonomic computing; data center; performance modelling; resource allocation

Abstract:
Distributed resource allocation in data centers has gained a lot of attention from the research community in the last few years, especially in fields like cloud computing and multitier systems. It is usually expected that these systems deliver some performance guarantees to users’ Service Level Agreements (SLAs). Therefore, data center servers may need to be dynamically redeployed to optimize some performance metrics so that to meet the promised SLAs. Moreover, the total profit of a system depends on its ability to reduce a data center’s energy cost through the resources utilization optimization. The main challenge of resource allocation is then to find the minimum amount of resources that an application needs to meet the desired Quality of Service (QoS). In this direction, autonomic computing appears to be one of the most popular concepts to achieve these goals by means of self optimization. These properties provide a system with a dynamic optimization of its own resources use, and enable it to autonomously adapt itself to its environmental changes. However, such autonomic resource allocation strategies may result in a loss of performance or even service degradation under some conditions. Furthermore, it is interesting to predict the behaviour and the efficiency of those strategies, before applying a new resource allocation, to forecast the most appropriate configuration and ensure the effectiveness of the autonomic manager. Thus, we propose in this paper a general insight of performance modelling of resource allocation strategies using the modelling of an autonomic resource allocation server as an example. The modelling is based on stochastic Petri net models (SPN). We consider in our modelling dynamic allocation strategies, based on workload intensity and user mixes. Finally, we illustrate the effectiveness of our modelling through a set of experimental results.

Pages: 18 to 24

Copyright: Copyright (c) IARIA, 2014

Publication date: April 20, 2014

Published in: conference

ISSN: 2308-3913

ISBN: 978-1-61208-331-5

Location: Chamonix, France

Dates: from April 20, 2014 to April 24, 2014