Zusammenfassung
Die Arbeit behandelt die Leistungsbewertung von Massenspeichereinheiten, bestehend aus einer Anzahl von Magnetplatten mit unabhängig voneinander beweglichen Schreib-Leseköpfen, die alle mit demselben Ein-Ausgabekanal verbunden sind. Die Frage, die wir beantworten wollen, ist die folgende: Welche Leistungsverbesserung kann erwartet werden, wenn die simultane Suchtechnik angewendet wird, d.h. wenn die langsamen Suchoperationen auf den verschiedenen Platten überlappend durchgeführt werden? Zu diesem Zweck werden Markov-Modelle sowohl für den Fall ohne simultanes Suchen als auch für den Fall mit simultanem Suchen entwickelt. Diese Modelle werden unter der Annahme analysiert, daß ein stochastisches Gleichgewicht besteht. Von der erzeugenden Funktion der Systembelegung können wichtige Leistungsgrößen, wie Durchsatz und mittlere Systembelegung erhalten werden, teilweise als explizite Funktionen der Systemparameter, teilweise auf numerischem Weg.
Abstract
This paper considers the performance evaluation of a mass storage device, composed of a number of independent movable head magnetic disk units, which are all connected to the same I/O-channel. The question we want to answer is the following: what performance improvement can we expect if we apply the concurrent seek technique to this device, i.e., if we allow the (slow) seekoperations on different disk units to overlap? To that end Markov chain models are developed both for the situation without concurrent seeks and the situation with concurrent seeks. These models are analyzed under the assumption that a stochastic equilibrium is achieved. From the probability generating function of the system occupancy important performance measures, such as throughput and mean system occupancy are obtained, partially as explicit functions of the system parameters, partially in a numerical way.
A quantitative and qualitative comparison between the different approaches is made.
This is a preview of subscription content, log in via an institution to check access.
References
Abate, J., H. Dubner, andS.B. Weinberg: Queueing analysis of the IBM 2314 disk storage facility. J. ACM15, 1968, 577–588.
Browne, J.C., et al.: The effect on throughput of multiprocessing in a multiprogramming environment. IEEE Trans. on Comp.22, 1973, 728–735.
Bruneel, H.: Analysis of single- and double-channel I/O-configurations. Research report ICI, Ghent State University, Ghent 1980.
Chow, Y.-C., andW.H. Kohler: Models for dynamic load balancing in a heterogeneous multiple processor system. IEEE Trans. on Comp.28, 1979, 354–361.
Davenport, W.B.: Probability and random processes. An introduction for applied scientists and engineers. New York 1970.
Fife, D. W., andJ.L. Smith: Transmission capacity of disk storage systems with concurrent arm positioning. IEEE Trans. on Comp.14, 1965, 575–582.
Foster, F.G., andH.G. Perros: On the blocking process in queue networks. Europ. J. Operational Res.5, 1980, 276–283.
Franklin, M., andA. Sen: An analytic response time model for single- and dual-density disk systems. IEEE Trans. on Comp.23, 1974, 1269–1276.
Gotlieb, C.C., andG.H. MacEwen: Performance of movable-head disk storage devices. J. ACM20, 1973, 604–623.
Hine, J.H., I. Mitrani, andS. Tsur: The control of response times in multi-class systems by memory allocation. Comm. ACM22, 1979, 415–424.
Kleinrock, L.: Queueing Systems. Volume I: Theory. New York 1975.
Little, J.D.C.: A proof for the queueing formulaL=λW. Operations Research9, 1961, 383–387.
Mehra, S.K., J.W. Wong, andJ.C. Majithia: A comparative study of some two-processor organizations. IEEE Trans. on Comp.29, 1980, 44–49.
Oney, W.C.: Queueing analysis of the Scan policy for moving-head disks. J. ACM22, 1975, 397–412.
Perros, H.G.: Modelling a disc I/O system: A queue network approach. Europ. J. Operational Res.1, 1977, 139–145.
Smith, A.J.: An analytic and experimental study of multiple channel controllers. IEEE Trans. on Comp.27, 1979, 38–49.
Teorey, T.J., andT.B. Pinkerton: A comparative analysis of disk scheduling policies. Comm. ACM15, 1972, 177–184.
Tou, J.T.: Digital and sampled-data control systems. New York 1959.
Wilhelm, N.C.: An anomaly in disk scheduling: a comparison of FCFS and SSTF seek scheduling using an empirical model for disk accesses. Comm. ACM19, 1976, 13–17.
Wilhelm, N.C.: A general model for the performance of disk systems. J. ACM24, 1977, 14–31.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bruneel, H. The application of the concurrent seek technique in order to improve the performance of disk storage devices. Zeitschrift für Operations Research 27, B113–B140 (1983). https://doi.org/10.1007/BF01916934
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF01916934