Re: Why MSDP?

[John Meylor <jmeylor@cisco.com>]

Vijay Gill wrote:
> 
> On Thu, 22 Mar 2001, Bill Fenner wrote:
> 
> >
> > One reason: the timescales for change of active source indications are
> > much different than BGP was designed to carry.  BGP wants to carry
> > data that doesn't change very often, e.g. see route dampening.  Sources
> > can come and go at an arbitrary rate, so the rate of change of the
> > information is potentially much higher.
> 
> So churn wasn't ok for BGP, but its ok in MSDP? Just wondering.

Vijay, yes,
sa turnover and cache size variability in MSDP is 
perfectly normal and expected behavior in 
the interdomain msdp topology, (reflecting sources
becoming active or non-active which is cool)
and with basic filtering in place (for bogus/unwanted
registers, sa-messages, etc) this "churn" is ok.  
 (Of course in BGP you don't expect lots and lots
of prefixes to be coming and going regularly, so
thats not ok.)
 However, just as with unicast reachability in BGP, 
if you don't control what information you are receiving
or announcing (eg in bgp from redistributing igp, or not filtering 
on peering), then msdp churn can quickly become "not ok".
 I think one of the conclusions drawn at MBONED WG
this week was to define some standard policy practices
for PIM/MSDP which could help make sure we can avoid
unwanted contributions to churn in msdp topology, see
Marshall Eubanks talk.
 

 John