Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
 

Traffic Matrix Solver Overview

In your network model, a set of end-to-end demands/flows is needed to perform various design and simulation studies. A few sources, such as Cisco’s NetFlow/TMS, Juniper’s JFlow, LDP traffic statistics, and LSP tunnel traffic statistics from SNMP, can provide end-to-end traffic information. However, this is usually CPU intensive, so the data is often partial. Most traffic collection systems, including MRTG, Infovista, and Concord eHealth, and NorthStar Planner’s traffic collector, provide interface traffic information. If you only have access to interface traffic data and/or partial end-to-end flow traffic data, you can still derive a reasonable set of end-to-end demands using the NorthStar Planner Traffic Matrix Solver.

Note:

The NorthStar Planner Traffic Matrix Solver addresses the following problem:

Given (a) the interface traffic utilizations in the network, (b) an optional trafficload file defining the bandwidth for a subset of the flows in the network, and (c) a set of flows indicating the sources and sinks in the network, determine the bandwidth of these flows to produce the given interface traffic utilization values.

This problem has no one right answer. Mathematically, it has infinitely many solutions. However, by supplying a little extra information, you can influence the NorthStar Planner Traffic Matrix solver to choose a solution that better fits the characteristics of your network. For example, you can indicate which nodes are sources and sinks of traffic (e.g., edge nodes). The remaining transit nodes will be limited to carrying “pass-through” traffic.

Once a possible traffic matrix solution has been derived, you can perform numerous traffic engineering studies. For example, you can run simulations to study whether the traffic flows can be rerouted safely during network failures. Or, you can use NorthStar Planner’s design capabilities to determine how to optimize cost and reliability for the given traffic. You may have collected interface utilization data for multiple periods. For each period, you can compute a set of end-to-end demands, especially times with heavy usage. Using this data, you can begin to build a picture of how your network traffic changes over time.