[BBLISA] BGP and multicast (thread renamed)

[Dean Anderson]

I personally have not proposed anything for equal cost routes. But
others have, and that runs on the routers.  Multipath has been around
quite a while and is implemented by pretty much all routers.

You should really be looking at RFC1812 and of course the BGP multipath
enhancements if you want to know how routing works on real routers.  
Some of the "clarification" parts of RFC's like 4271 have largely been
ignored by vendors, and are contrary to e.g. multipath. For example,
4271 doesn't describe multipath at all.

I also observe that the quality of work out of the IETF has greatly
dropped off in the last 10 years or so due to many factors not least of
which is lack of participation (e.g the IETF main list now has only 1700
members; compare with the very obscure MVS 3.8J list which has over 2500
members.)  You almost certainly have never heard of MVS3.8J;  It is an
ancient version of MVS from 1978. There are more people interested in
hacking an ancient mainframe operating system than participating in the
IETF.  Probably better known to bblisa is the comparable 1978 pre-sysV
unix code. Anyway, the relevance of some new RFCs to running code
reflects the growing irrelevance of the IETF standards, especially some
new standards completely contrary to what anyone runs or wants to run.  
Of course not everything from the IETF is crap, but its getting spotty.

		--Dean

On 21 Jul 2010, Daniel Hagerty wrote:

> Dean Anderson <dean at av8.com> writes:
> 
> > However, both A and B _can_ have equal cost routes to the D. The more
> > anycast instances of D you have, the more likely that A and B will both
> > have equal cost routes to D
> 
>     What mechanism do you propse for this "equal" thing?  BGP's route
> comparator doesn't have "equal".  rfc4271, 9.1.2.2, bullet g.
> 
> 
> 
> 

-- 
Av8 Internet   Prepared to pay a premium for better service?
www.av8.net         faster, more reliable, better service
617 256 5494