Browse Prior Art Database

Heuristic Method for Grouping Based on Traffic Counts

IP.com Disclosure Number: IPCOM000110609D
Original Publication Date: 1992-Dec-01
Included in the Prior Art Database: 2005-Mar-25
Document File: 2 page(s) / 73K

Publishing Venue

IBM

Related People

Chang, DY: AUTHOR [+2]

Abstract

Disclosed is an effective and efficient heuristic method for grouping nodes in a massive-node server system based on traffic counts between nodes. For a group-based massive-node server status monitor, the information of the nodes are presented in a group level. The user may be interested in various ways of grouping nodes. Traffic counts between nodes are usually used as a base for grouping. The other attributes used as a base for grouping have association with one node only. Traffic counts have more than a single node involved. In addition, each node has traffic counts with multiple nodes. This complicates the grouping method based on traffic counts. Simple algebraic calculation is not sufficient.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Heuristic Method for Grouping Based on Traffic Counts

       Disclosed is an effective and efficient heuristic method
for grouping nodes in a massive-node server system based on traffic
counts between nodes. For a group-based massive-node server status
monitor, the information of the nodes are presented in a group level.
The user may be interested in various ways of grouping nodes.
Traffic counts between nodes are usually used as a base for grouping.
The other attributes used as a base for grouping have association
with one node only.  Traffic counts have more than a single node
involved.  In addition, each node has traffic counts with multiple
nodes.  This complicates the grouping method based on traffic counts.
Simple algebraic calculation is not sufficient.

      For a massive-node server system, the basic principle of
grouping based on traffic counts between nodes is to put the nodes
with high traffic counts into the same group.  This heuristic method
considers the traffic count between two nodes as the similarity
measurement of these two nodes.  Similar nodes attract each other to
merge into a group.  Groups with high similarity also attract each
other to merge into one group.  The similarity between a node N and a
group of nodes G is measured by the total traffic count between N and
G.  By extending this measurement, the similarity between two groups
of nodes is the average value of traffic counts among the pairs of
nodes, one from each group, respectively.  By extending the concept
from node-level to group-level, this heuristic method is described as
follows.
  (1) Predetermine the maximum number of nodes allowed in a group (M
sub #no...