Hybrid Analysis/Simulation: Distributed Networks
Original Publication Date: 1976-Jun-15
Included in the Prior Art Database: 2007-Mar-30
Software Patent Institute
Sauer, C.H.: AUTHOR [+4]
AbstractRC 6341 (#26178) 6/15/76 (Rec'd 12/30/76) Computer Science 1 2 pages Hybrid Analysis/Sirndation: Distributed Networks
RC 6341 (#26178) 6/15/76 (Rec'd 12/30/76) Computer Science 1 2 pages
Hybrid Analysis/Sirndation: Distributed Networks
Sauer, L.S. Woo W. Chang
IBM Thomas J. Watson Research Center IBM Data Processing Division Yorktown Heights, New York 10598 White Plains, New York 10604
Abstract: Hybrid solution techniques combining analysis and simulation can be used for complex models when analytic solutions are intractable and simulation solutions are unwieldy. Models of distributed teleprocessing networks are often complex since processing capability is distributed at several nodes instead of having all processing done at a host computer. A model has been developed to study some of the characteristics of the Systems Network Architecture (SNA), Level 2, and related software. This model is solved using hybrid techniques.
LIJIITED DISTRIBUTION NOTICE
This report has been submitted for publication elsewhere and has been issued as a Research Repoe for early dissemination
of its contents. ,As a courtesy to the intended publisher, it should not be widely distributed until after the date of outside publication.
Copies may be requested from:
0331 Thomas J. Watson Research Center Post Office Box 218
Yorktown Hetghts, Sew York 10598
The primary purpose of this paper is to report on preliminary efforts at combining analytic and simulative methods to develop performance evaluation approaches which will be appropriate to configuration of current and future distributed networks. We refer to this approach as a hybrid of analyis and simulation, or briefly, as hybrid simulation. In the remainder of the paper, we will mainly discuss those aspects of distributed networks which we believe cannot be dealt with analytically. We will characterize these aspects in more detail. Then we will discuss how they may be dealt with by hybrid simulation. Finally, we discuss results from a sample system model which we have constructed.
Teleprocessing networks have allowed many computer users to simultaneously have individual access to computers. One approach to teleprocessing is the centralized network, where a central computer.system is directly connected to terminals by communication links. The central system is responsible for all of the user computation and is also responsible for handling the communication links and terminals. The centralized approach has several problems:
1. The system may spend a considerable portion of its resources on communicating messages associated with minor user activities.
2. The major portion of the response time seen by minor user activities will be the time messages spend in the communication links.
3. If the central computer fails, then all of the terminals fail.
4. The communication l...