STOCHASTIC MODELS OF MULTIPLE AND TIME-SHARED COMPUTER OPERATIONS.
Abstract
Currently, both the hardware and software designs of many large computing systems aim at two basic objectives: improved system performance through exploitation of parallelism in multiprocessor systems, and the provision of improved man-machine interation primarily through time-shared computer systems. In studying these systems mathematical modeling and analysis constitute an important step towards providing design tools that can be used in building such systems. In the work being described here queueing models of these types of systems are presented and analyzed with the purpose of obtaining conventional performance measures such as the steady-state probability distributions of the number in the system and waiting time in the queue. In particular, a multiprocessor queueing model is devised in which arriving jobs consist of a random number of segments operable in parallel. This model may be viewed as a type of bulk arrival, multiple channel queueing model with exponential service and inter-arrival times. The performance measures mentioned above are obtained and compared with those resulting from other, similar types of computer models in which parallel-processing capabilities are constrained or nonexistent.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1966
- Accession Number
- AD0636976
Entities
People
- Edward C. Coffman
Organizations
- University of California, Los Angeles