MIT Department of Electrical Engineering & Computer Science

Group Mutual Exclusion

Yuh-Jzer Joung

Friday, October 22, 1999
1:00 PM
NE43-516
Theory of Distributed Systems Seminars

Abstract

Group mutual exclusion occurs naturally in situations where a resource can be shared by processes of the same group, but not by processes of different groups. For example, suppose data is stored in a secondary memory device (such as a CD jukebox). Since only one disk can be loaded for data access at a time, when a disk is loaded, users that need data on this disk can concurrently share the disk, while users that need a different disk have to wait until no one is using the current loaded disk. As can be seen, group mutual exclusion is more general than the classical n-process mutual exclusion problem and the readers and writers problem. Although hundreds of papers have been devoted to the two classical problems in the last three decades or so, group mutual exclusion has not been addressed in the literature until 1998.

In this talk, I will discuss the design issues concerning group mutual exclusion, present solutions for the problem in both shared memory and message passing models, and, most importantly, propose several directions for future work.

Reference: 1. Asynchronous Group Mutual Exclusion, Yuh-Jzer Joung, PODC 98. 2. The Congenial Talking Philosophers Problem in Computer Networks, Yuh-Jzer Joung, DISC 99.

(available in the TDS wall pockets outside NE43-366)