Augmented CPU reservations: towards predictable execution on general-purpose operating systems

Update Item Information
Publication Type Journal Article
School or College College of Engineering
Department Computing, School of
Creator Regehr, John
Other Author Stankovic, John A.
Title Augmented CPU reservations: towards predictable execution on general-purpose operating systems
Date 2001-01-01
Description One problem with performing soft real-time computations on general-purpose operating systems is that these OSs may spend significant amounts of time in the kernel instead of performing work on behalf of the application that is nominally scheduled: the OS effectively steals time from the running application. Stolen time can be a significant obstacle to predictable program execution on real-time versions of Linux and Windows 2000, where it can cause applications to miss essentially all of their deadlines. We propose augmented CPU reservations, a novel mechanism for using fine-grained accounting information about the amount of stolen time to help the scheduler allow applications to meet their deadlines. We have designed and implemented Rez-C and Rez-FB, two schedulers that provide augmented reservations, and we have tested them in Windows 2000, showing that they can increase the predictability of CPU reservations. We also experimentally quantify the severity of stolen time caused by a variety of devices such as hard disk controllers, a network interface, and a software modem under real-time versions of Windows 2000 and Linux.
Type Text
Publisher Institute of Electrical and Electronics Engineers (IEEE)
First Page 1
Last Page 8
Dissertation Institution University of Utah
Language eng
Bibliographic Citation Regehr, J., & Lepreau, J. (2001). Augmented CPU reservations: towards predictable execution on general-purpose operating systems. In Proceedings of the 7th Real-Time Technology and Applications Symposium (RTAS 2001), 1-8. May30-June1.
Rights Management (c) 2001 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Format Medium application/pdf
Format Extent 113,594 bytes
Identifier uspace,17512
ARK ark:/87278/s6h13krj
Setname ir_uspace
ID 707980
Reference URL https://collections.lib.utah.edu/ark:/87278/s6h13krj