Departament de Cičncies de la Computaciķ
Home Spanish  |  English

Id. report:  LSI-10-21-R
Title:  An Integer Linear Programming Representation for DataCenter Power-Aware Management
Author(s):  Berral, J. LL.,  Gavaldā, R.,  Torres, J.
Date (year-month-day):  2010-11-17
Document Type: 
Keyword(s):  Integer Linear Programming, Autonomic Computing, Scheduling, Modeling
Abstract:  Nowadays energy-related costs have become one of the major economical factors in IT data-centers, and companies and the research community are currently working in new efficient power-aware resource management strategies, also known as “Green IT”. But despite power awareness, the classical factors like economical benefit and quality of service are still essentials for the correct performing of the Cloud businesses. So efficiency must meet power saving, economic benefit and client satisfaction, points often all-three incompatible. This work exposes how to represent a grid data-center based scheduling problem, taking the advantages of the virtualization and consolidation techniques, as a linear integer programming problem including all three mentioned factors. Although being integer linear programming (ILP) a computationally hard problem, specifying correctly its constraints and optimization function can contribute to find integer optimal solutions in relative short time. So ILP solutions can help designers and system managers not only to apply them to schedulers but also to create new heuristics and holistic functions that approximate well to the optimal solutions in a quicker way. The proposed model is evaluated comparing the optimization results in an only power-aware scheduling, where the system minimizes the power consumption via consolidation; a revenue-based scheduling, where the system maximizes the benefit of running machines against the power consumption; and a quality of service oriented scheduling, where some relaxation towards QoS is tolerated in exchange of avoiding expensive operations. Results demonstrate that ILP is useful in order to find good schedules and the lower bounds for the scheduling optimization, so new heuristics and policies can be obtained by weighting the involved elements (revenue, power, quality of service) and observing the system behavior.
Document properties 
Number of pages:  21
Size:  346.18 KB
File type:  pdf  (Zipped)