Load Sharing in a Computer-Communication Network

Abstract

This study investigates load sharing in a system of computers interconnected by a store and forward communication network. The problem is analyzed by modeling both computers and communication channels as queues and evaluating system performance on the basis of the steady state expected time to process computer jobs in the system. Upper and lower bounds on this performance criteria are developed and used to define regions of operation for a network using load sharing. Two techniques for load sharing are then presented. The first technique, called statistical load sharing, consists of sending a fraction of the jobs arriving at overloaded computers to underloaded computers by random sampling. This technique is analyzed by a network of queues model. It is shown that the general formulation of statistical load sharing is a nonlinear multicommodity flow problem which can be solved by an efficient computer algorithm. The improvement is system reliability due to the ability of load sharing to provide emergency backup in case of computer failure is also studied. The second technique for load sharing, a type of dynamic load sharing, makes job assignments to computers on the basis of the computers not busy at the time of assignment. This technique is analyzed by an approximation to the hypercube queueing model.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1976
Accession Number
ADA032135

Entities

People

  • Eberhard Frank Wunderlich

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Communication Channels
  • Communication Systems
  • Computations
  • Computer Communications
  • Computer Networks
  • Computer Programs
  • Computers
  • Mathematical Models
  • Mesh Networks
  • Operations Research
  • Probability
  • Random Variables
  • Reliability
  • Ring Networks
  • Statistical Sampling
  • Steady State

Fields of Study

  • Computer science
  • Engineering

Readers

  • Computer Networking
  • Mathematical Modeling and Probability Theory.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.