BGP4+

[bound@zk3.dec.com]

Pedro,

>There are two BGP proposals for IPv6:

>- draft-bates-bgp4-multiprotocol-01.txt,  a.k.a. BGP 4+.
>- draft-stewart-bgp-multiprotocol-00.txt, a.k.a. BGP 5.
>
>I don't think there is any possible confusion on the term BGP 4+.

I know.  But all I saw was BGP4 and did not differentiate the 5.  I also
heard a rumor there was a compromise draft of BGP4 btw the two drafts
above.  Guess not.

    Jim> Can someone from Cisco share with us is this the draft?

>I'm not speaking for cisco. Just giving you an overview of what are the
>proposals at the table in the IDR WG.
>Also i'm curious why you asked explicitly about cisco and not telebit.

I did not see where Telebit implemented BGP on the 6bone?  

>Some people believe that interoperability testing on the 6bone is more
>valuable than UNH testing. The whole purpose of the 6bone is
>interoperability testing.

Well they don't know what they are talking about.  And the Linux comment
is bogus we are working on getting Linux in the next test period at UNH
as a University implementation which does not have to pay the fee.  
Only vendors, research institutes,  and consortias are charged.

The test suites at UNH test specific features and will cause
ill-behavior to see if the implementation does the right thing according
to the spec. There are a specific set of tests of the assertions from the
specifications for Hosts and Routers.  The 6bone is a good test method
for parts of IPv6.  But we need more than that to test IPv6.     

>But i'm not sure i understand your message. Do you propose that UNH issues
>certifications for ipv6 implementations and that non certified implementations
>would not be allowed in the 6bone and/or cairn ?

ABSOULETLY NOT.  Anyone who would suggest such a thing is a moron.  I
have no clue how you came up with this one at all. 

>I fear that such meassure would left off some of the most popular unix
>implementations (e.g. Linux) from the 6bone. Also it is not clear how do
>you propose to enforce such restriction.

I am not so have no response to this.

    Jim> I would like to see in the RIPE-CC or somewhere a connotation
    Jim> who has done interoperability testing and who has not.

>I'm more confused. RIPE-NCC ftp site has a list of 6bone nodes. You mean the 
>6bone sites should participate in UNH testing ? Or should the registry have
>a list of blessed implementations ?

I am saying any implementation on the 6bone by definition has done
testing.  This is idempotent.  

In addition have:

  Participated in Connectathon:  YES or NO
  Participated in UNH Testing :  YES or NO

That is all.  Whether we like it or not the market is watching the 6bone
and if you have done the above you have tested your implementation in 
another very rigorous interoperability environment.  Also at least 90% of the
implementations have (inluding Cisco) been at UNH, also Netbsd and Freebsd 
has done this too.  So I think its a positive entry in the RIPE to add to the
demonstration that the 6bone nodes are serious implementations.  Its
not a negative thing but a positive thing I propose.  

Regarding Linux.  I did a talk at UNH to the NH Linux Users Group on
IPv6 about 7 weeks ago.  From that meeting I think we determined we
need to get Linux as status quo at the next UNH Test Period.  I am
working that in my personal time to make that happen (not Digital time).
I am doing an IPv6 dinner talk in July for the Boston Linux and Internet
SIG Group in July.  Again objective is to get Linux in the UNH test
loop.

Do you get the point now.  I care less if you agree but you came up with
all these negatives.  Or are you just mistrust of me personally and
picking a fight with me?   If thats the case lets take it off line and I
will get into a flame fest with you privately.   

/jim