Determining Accurate Traffic

Matrix for IP Networks

Based on NANOG39: Best Practices
for Determine the Traffic Matrix in IP
Network (Thomas Telkamp)
Presented by: Femi Adelani
ECE Department, University of Manitoba

Background

Traffic Matrix: the amount of data

transmitted between every pair of
network nodes
Traffic Matrix can represent peak
traffic, or traffic at a specific time
Router-level or PoP-level matrices

Purpose

Analysis and Evaluation of other network states than

the current
Capacity Planning

Resilience Analysis

network changes
what-if scenarios
Could be per class
network under failure conditions

Optimization

Topology Find bottlenecks

OSPF/IS-IS metric optimization/TE
MPLS TE tunnel placement
3

Types of Traffic Matrices

Internal Traffic Matrix

PoP to PoP matrix

Can be from core (CR) or access (AR) routers
Class based

External Traffic Matrix

PoP to External AS
BGP
Origin-AS or Peer-AS
Peer-AS sufficient for Capacity Planning and Resilience
Analysis
Useful for analyzing the impact of external failures on the core
network (capacity/resilience)

Traffic Matrix Data Collection

Data is collected at fixed intervals

Measurement of Byte Counters

E.g. every 5 or 15 minutes

Need to convert to rates
Based on measurement interval
Counter roll-over issues

Create Traffic Matrix

Peak Hour Matrix - 5 or 15 min. average at the

peak hour
Peak Matrix - Calculate the peak for every demand
5

Collection Methods

NetFlow

BGP Policy Accounting/DCU

RSVP - Measurement of Tunnel/LSP counters

LDP - Measurement of LDP counters

Estimation

Routers collect aggregated destination statistics

MPLS

Routers collect flow information

Export of raw or aggregated data

Estimate Traffic Matrix based on Link Utilizations

Issues associated with each method?

6

General Formulation
Y = AX
Y = Link

Utilizations

A = Routing

Matrix

X = Po int to Po int

Demands

Routing Matrix
644474448
Y1
X
a11 a12 . . a1k 1
X 2
Y2 a
a
a
. . 2 k
22
. 21

.
= .
.
.
.
.
.
.

.
. . . .
.
al1 al 2 . . alk .
Y
X
l
k

Demand Estimation

Underdetermined system:

Many algorithms exist:

N nodes in the network

O(N) links utilizations (known)
O(N2) demands (unknown)
Must add additional assumptions (information)
Gravity model
Iterative Proportional Fitting (Kruithofs Projection)
Maximum Likelihood Estimation
Entropy maximization
Bayesian statistics (model prior knowledge)
Etc...!

Calculate the most likely Traffic Matrix

Related Works

Several algorithms have been published

Comparisons and Improvements have
been proposed and implemented
Commercial Software e.g Cariden MATE
Software
Best Practices for Implementation have
been suggested
9

Looking Ahead

Issues with data collected and overhead

associated with the collection method Best
Practice
Accuracy Traffic Matrix Estimation based on
existing methods
Possibility of proposing a new TM Estimation
Method or improving on existing ones
Use proposed TM for Capacity planning,
Route Optimization and resilience analysis
using real data and measure performance
10

11