ISP Backbone Traffic Inference Methods to Support Traffic Engineering

12/19/00

Click here to start

Table of Contents

ISP Backbone Traffic Inference Methods to Support Traffic Engineering

Outline 

RATIONALE

Problem Description

Problem Description

Inputs to the model

DETERMINISTIC INFORMATION

MEASURED INFORMATION

USAGE INFORMATION

PPT Slide

PPT Slide

PPT Slide

PPT Slide

CONSTRAINTS

PPT Slide

PSEUDO-INVERSE METHOD

LINEAR PROGRAMMING METHOD

OBJECTIVE FUNCTION COEFFICIENTS

EXAMPLE

If all objective coefficients are equal to 1

But if coefficient are equal to the number of hops of demand route

Another advantage of the LP method

Test Results

Test Results

Objective coefficients = number of hops

Objective coefficients = number of hops

All objective coefficient = 1

Netflow turned on on five random routers

Netflow turned on on five most used routers

Netflow turned on on ten random routers

Comparison of different results

Conclusions

Author: Olivier Goldschmidt

Email: isma@caida.org

Home Page: http://www.caida.org/workshops/isma/0012/

Other information:
Internet Statistics and Metrics Analysis Workshop - Winter 2000 
San Diego Supercomputer Center, 
University of California, San Diego 
December 7-8, 2000 

Download presentation source