All things related to multicast

[John Kristoff <>]

Hello,

As part of the background for a recent presentation, I have a couple
of relatively simple tools I created to help understand multicast DoS
issues.  However, I've had a hard time finding people who can do some
extended testing and tell me what, if anything, these would do in a
real multicast-enabled internetwork.  I don't have a good lab setting
to really put this in and I'm too afraid to run these in my production
network.

I'm soliciting anyone who might be interested in trying these tools
out in their own isolated environment and reporting back, to me or
perhaps even this list if others do not find that objectionable, the
results.  I have two basic tools. One sends ICMP echo messages to each
group in 224/4 in turn.  The other sends an IGMPv2 join to each 224/4
group in turn.  Both operate in rapid fire mode, as fast as Perl can
run on your machine.  I'm too afraid to run these in my own network,
but I'm interested in seeing what happens with the following:

  Layer 2 switch IGMP snooping effects
  IGMP join state on the router interface effects
  PIM state at the DR and upstream, including the RP
  MSDP SA creation workload and flooding effects

I have some ideas for at least a couple of other DoS related tools
if anyone feels this development is worthwhile.  Of course, each of
these could be made a lot nastier if tuned and written in something
like C.

A hopefully obvious word of caution... these scripts could do some
serious damage if run on your production network and not only impact
your network, but other networks such as Abilene and my network.  I
expect anyone who expresses interest in this to have enough clue to
not cause you, me or anyone else any harm.

Contact me for details.

John