Browse Prior Art Database

Analyze Host/Network Traffic for Worst-Case Performance

IP.com Disclosure Number: IPCOM000060881D
Original Publication Date: 1986-May-01
Included in the Prior Art Database: 2005-Mar-09
Document File: 2 page(s) / 14K

Publishing Venue

IBM

Related People

Adolph, G: AUTHOR [+2]

Abstract

This article describes a method whereby network operations can focus on areas of worst-case performance within their network. The analysis and reporting of worst-case performance occurs on a real-time basis from information furnished by a traffic collection system. The traffic collection system used is the Network Performance Monitor (NPM). Network traffic analysis for worst-case performance can be directed to areas within a network or to the network in general. Once initiated, the analysis results are presented on an automatically refreshing basis so that performance trends can be observed. Network operation, charged with the smooth operation of sometimes very large networks, normally functions in a react mode with respect to network performance.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Analyze Host/Network Traffic for Worst-Case Performance

This article describes a method whereby network operations can focus on areas of worst-case performance within their network. The analysis and reporting of worst-case performance occurs on a real-time basis from information furnished by a traffic collection system. The traffic collection system used is the Network Performance Monitor (NPM). Network traffic analysis for worst-case performance can be directed to areas within a network or to the network in general. Once initiated, the analysis results are presented on an automatically refreshing basis so that performance trends can be observed. Network operation, charged with the smooth operation of sometimes very large networks, normally functions in a react mode with respect to network performance. Whether dealing with hundreds or tens of thousands of network resources, it is advantageous to obtain advance notice of potential problem areas. Without advance notice, network operations only become aware of problem areas by calls from users. The method described by this article provides advance notice of potential performance problem areas so that action can be taken before a situation of widespread user dissatisfaction occurs. Performance problems in this context are defined as follows: 1. Excessive traffic volumes - Networks are designed to handle an estimated traffic volume. The volume in terms of the number of characters and the number of transmission packets from/to a destination/origin is sized and network resources are set in place to handle that load. If the traffic volume exceeds the network design point, then queues begin to build and degradation of service occurs. 2. Operator response time problems - Response time problems are the overt manifestation of degradation of service. Response times vary due to flow state variations in the network. However, for any one transaction type, the response time should only vary within a small range. Consistent elongation of response time outside of that range is considered a response time problem. To do worse-case analysis, information on performance states must be obtained on a real-time basis. This may be done by an information collector such as NPM. When starting collection, a user may specify response time thresholds. These thresholds can then be used to d...