A Buffered Disk Subsystem
The paper presents an analysis of buffered disk subsystems. First, the waiting time to start the service of an IO request is modelled. The model shows that the waiting time is negligible. This is a motivation to model a buffered disk subsystem. The average response time in such a subsystem can be reduced by 50%, and throughput can be doubled in comparison to a standard subsystem.
Key wordsdisk subsystem buffered disks performance analysis queueing modeling Markov process
Unable to display preview. Download preview PDF.
- [Bran81]Brandwajn A., “Multiple Paths Versus Memory for Improving DASD Subsystem. Performance”, Performance’81, F.J. Kylstra (ed.), North-Holland, 1981, 415–434Google Scholar
- [GGMW91]Grygiel K., Gruilewski T., Mincer-Daszkiewicz J., Weiss Z., “Evalnet: A Step Towards Integrated Environment for Computer System Modelling”, Fifth International Conference on Modelling Techniques and Tools for Computer Performance Evaluation, Torino, Italy, February 13–15, 1991Google Scholar
- [Hout85a]Houtekamer G., “Performance Analysis of the IBM DASD Subsystem, and Some Suggestion for Improved Systems”, Proceedings of 3 GI/NTG Conference, Dortmund, West Germany, October 1985, 261–276Google Scholar
- [Hout85b]Houtekamer G., “The Local Disk Controller”, Comm. ACM 28 (1985), 173–182Google Scholar
- [KrHa90]Krzesinski A.E., Hamse H.S., “Performance Models of IBM DASD Subsystems”, Performance’90, Edinburgh, Scotland, September 1990Google Scholar
- [Lave83]Lavenberg S.S. (ed.), Computer Performance Modelling Handbook, Academic Press, New York, 1983Google Scholar
- [NaBa89]Narasimha Reddy A.L., Banerjee P., “Performance Evaluation of Multiple-Disk I/O Systems”, IEEE Transactions on Computers, 1989Google Scholar
- [Teun91]Teunissen P., “Cache Modelling for IBM DASD”, Fifth International Conference on Modelling Techniques and Tools for Computer Performance Evaluation, Torino, Italy, February 13–15, 1991Google Scholar