Control and modeling issues in computer operating systems: resource management for real-rate computer applications
- 11 November 2002
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3 (01912216), 2212-2221
- https://doi.org/10.1109/cdc.2000.914125
Abstract
Commonplace computer applications on general-purpose computers increasingly are expected to meet "real-rate" requirements, processing or displaying data or images at an externally driven "rate". We describe a feedback-control-based resource manager design approach, allowing the computer system to allocate resources such as CPU and network bandwidth based on the measured "progress" of the applications. Progress is measured by separating a complex application into a number of simpler applications separated by buffers. The resource scheduler measures the buffer fill levels to determine whether the rates of data coming in and going out of each buffer are matched. Feedback controllers keep the buffer levels around a certain fill level. We have developed prototype systems in the Linux environment that demonstrate that (classical) feedback control can be used to match the real rates. However, more formal methods, such as those that can be developed by the control theory community, are needed to help with the analysis and design of such systems to make them commercially viable. This paper presents the computer system problems, results from the prototype designs showing feasibility, some preliminary modeling results, and demonstrations and discussions of which control modeling, analysis and design results and techniques appear to be relevant to this computer system problem, and why.Keywords
This publication has 5 references indexed in Scilit:
- Design and evaluation of a feedback control EDF scheduling algorithmPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Modeling and performance control of Internet serversPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Control challenges in multi-level adaptive video streamingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Numerical Methods for Optimal Control Problems with State ConstraintsLecture Notes in Mathematics, 1999
- Scheduling Algorithms for Multiprogramming in a Hard-Real-Time EnvironmentJournal of the ACM, 1973