Friedrich-Alexander-Universität UnivisSuche FAU-Logo
Techn. Fakultät Willkommen am Institut für Informatik FAU-Logo
Logo I4
Computer Science 4
Operating Systems
Distributed Systems
Research Projects
Student Projects
Phone 09131/85-28820
Fax 09131/85-28732

University of Erlangen
Computer Science 4
Dr. Frank Bellosa
Martensstraße 1
91058 Erlangen

Room 0.037
Department of Computer Science  > Computer Science 4  > Frank Bellosa  > Student Projects
Accounting and Control of Power Consumption in Energy-Aware Operating Systems
Martin Waitz
Advisor: Andreas Weißel, Dr.-Ing. F. Bellosa
Registered as Diplomarbeit SA-I4-2002-14 , Januar 31 2003
[Abstract] [Full Paper (pdf) , 363 kB]
An important task of operating systems is to schedule shared resources fairly between several parties. Precise accounting of consumed resources is the key to that goal. However, most operating systems only use very basic accounting strategies.

This thesis discusses methods for resource accounting and introduces a powerful, yet easy to use accounting model based on resource containers. The goal of this model is to always charge the party that is responsible for some resource usage. To achieve this, client-server relationships between running processes are detected. They provide an invaluable source of information which can be used to identify the entities initiating resource intensive actions. As the resource containers which are used for accounting can be nested to form a hierarchy, sophisticated accounting and scheduling policies can be formulated.

Utilizing this new accounting model, the energy consumption of the machine can be charged to the responsible entity. Accounting energy consumption is a very natural way, as every hardware component that contributes to the execution of a program is consuming energy. Several methods used to measure or estimate the energy consumption of various hardware parts are discussed, paying special attention to the main processor.

Peak energy consumption is especially important as many components have to be dimensioned according to the maximum power consumption. Reducing this peak consumption can save costs in both high-end data centers and small mobile devices. A software method for limiting energy consumption is introduced. By using the new resource model, advanced policies can be defined that allow one to control power consumption of the entire machine, individual processes or special clients and servers.

About our Server | Mail to Webmaster
Last modification: 2/10/2005