Testing and Verifying an IPv6 Based Multicast Network · PDF fileTesting and Verifying an IPv6 Based Multicast Network ... Vilmos Bilicki Testing and Verifying an IPv6-Based Multicast

MotivationOur Contribution

Summary

Testing and Verifying an IPv6 Based MulticastNetwork

Vilmos Bilicki

University of SzegedInstitute of Informatics

Department of Software Engineering

IPv6TD, 2006

Vilmos Bilicki Testing and Verifying an IPv6-Based Multicast Network


Summary

Outline

1 MotivationCurrent trendsIssues with the streaming contentFree tools for IPv6 multicast monitoring/testing

2 Our ContributionThe NetSpotter FrameworkMeasurements and Results

3 Summary



Summary

Current trendsIssues with the streaming contentFree tools for IPv6 multicast monitoring/testing

Outline



3 Summary



Summary


Current trends

There is an evolution of the network infrastructure and ofthe services provided

Broadband access is becoming more common{ 70% of total US homes}This will push the providers towards Triple Play services

Streaming content (TV,...)On-line communication (VoIP, Video Conferencing,...)Data access (Web, Email, ...)

IPTV is the most promising new serviceDigital high quality interactive content (e-voting, sceneselection, ...)Mobile access (if there is no DVB-H,...)On-line video stores



Summary


Outline



3 Summary



Summary


Issues with the streaming content

Problems to be solvedVery high number of usersNew user expectations / traffic types (delay, jitter, error rate)

Current solutions:Unicast communication -> inefficient bandwidth usageMulticast communication -> new technology (the same ageas the Internet :) )

IPv4 small address spaceLack of working experience in this fieldLack of testing and monitoring solutions in this field

Questions:Is my network ready for IPv6 multicast?What kind of QoS properties can I guarantee for mycustomers?How can I deal with special situations?



Summary


Outline



3 Summary



Summary


Free tools for IPv6 multicast monitoring/testing

For monitoring:

DBeacon?

TestingTTCN3 based toolsD-ITG?



Summary

The NetSpotter FrameworkMeasurements and Results

Outline



3 Summary



Summary


The NetSpotter framework

The goal of this framework is to provide an easy-to-useopen source tool for system administrators not just forsystem monitoring but for system testing too.The primary goal was to support the monitoring and testingof the IPv6 multicast networkIt provides the following set of services:

Network discovery (SNMP, L2, L3, MCast topology)Collecting different performance countersArbitrary traffic definitionMulti-point traffic generation and measuring

It is implemented in Java (J2EE and J2SE)



Summary


High level overview



Summary


Network testing capabilities

Traffic classesFrame levelPacket levelSocket level

Message constructionXML-based solutionInheritance is supportedDynamic values

Sequence constructionMSC-based solutionThe most important MSC features can be used (loop,reference,...)Synchronization can be achieved using:

Delay elementsSynchronizing messages from the central cluster



Summary


Work flow



Summary


Main screen



Summary


Sequence construction



Summary


Message construction



Summary


Outline



3 Summary



Summary


IPv6 multicast measurmenets

RFC 2544, IPv6 unicast Benchmarking Methodology draftSingle-port / Multi-portThroughputLatencyFrame loss...

RFC 3918The most interesting measurements:

Scaled Group MatrixGroup Join/Leave Delay

Main deficiencies of this RFC:It is only concerned with multicast forwarding not multicastroutingOnly ASM features are tested; no FSM or SSM testing isdefined



Summary


Our measurement

We measured the effects of a large number of channelsand sources on the forwarding capabilities of a Linux IPv6multicast router



Summary


Details

Testing platform:PC P4, 1300 MHz, 512 MByteDebian Sarge, MRD6 0.9.5 as a PIM-SM routerZebra Ripng as unicast routing algorithm

The measurementWe injected 50000 IPv6 multicast UDP packets from thesource agentPacket loss: the destination agent measured the number ofreceived packetsDelay: The destination joined a defined number of channelsand measured the time needed for receiving the subscribedchannels



Summary


Results

Without some kind of ACL list the multicast network isvulnerable to DOS attacksThe number of channels influences the delay and thepacket loss significantly

N.Ch 64 512 150010 50000 50000 49200100 49514 49664 43311

1000 46813 43808 4164210000 n.a n.a n.a60000 n.a n.a n.a

Table: Packet loss

N.Ch 64 512 150010 17 23 14100 227 254 319

1000 3800 3700 420010000 72777 >70000 >7000060000 >70000 >70000 >70000

Table: Delay (ms)



Summary

Conclusions

NetSpotter frameworkWe presented the NetSpotter framework which can be usednot just for protocol validation but network testing too

Measurements

We measured the MRD6 PIM-SM implementationIt turned out that one can easily initiate DOS attacksagainst the multicast network with several special packages



Summary

Future plans

The first version of the framework will be finished byautumn of this yearWe would like to create protocol templates for the mostimportant protocols (e.g. SIP, http, ...)In the second version we plan to do the following:

We would like to improve the defining and handling ofTTCN3-like logic formulasWe would like to extend the network discoveringcapabilities with L2 information (CAM,...) and with muliticastinformation from the agents



Summary

Thank you for listening

The homepage of the NetSpotter project:http://sourceforge.net/netspotter

Q and A


Documents

Testing and Verifying an IPv6 Based Multicast Network · PDF fileTesting and Verifying an IPv6 Based Multicast Network ... Vilmos Bilicki Testing and Verifying an IPv6-Based Multicast