Re: RFC1863 to historic? [Re: [Idr] Last Call on draft-ietf-idr-rfc2796bis-00.txt (fwd)]

At 7:48 AM -0400 5/4/04, Tulip Rasputin wrote:
>Jeff,
>
>[..]
>>  All of this said, there are other small exchanges that are said to
>>  be deploying route servers, but it sounds like these are small regionals.
>
>I would not agree to this! There is this european internet exchange 
>association (euro-ix) where
>europe's leading ISPs connect to permit the exchange of network 
>information. I am aware of most of
>them using route servers (though not the ones like defined in 1863).

Does Euro-IX have any documentation on the route servers it does use?

It sounds to me as if 1863 itself may need to be made Historic, but 
there may be a need for several new, more focused documents:

     1. Use of route servers for information collection.  This might actually
        be a subset of 1863, but perhaps with updates or crossreferences to
        such things as the new data collection communities.

     2. BCP for production use such as Euro-IX.

     3. Possibly something informational about L3 exchanges where
        an exchange-operated router is there primarily for routing
        information exchange. ISTR a Cisco workshop describing such
        techniques as a first step to high-capacity L2 exchanges, and
        I've seen this done both in small North American exchanges and
        reports on Asia-Pacific initial exchanges.

As an aside, this does get into the increasingly blurred area about 
how much the IETF RFC process is appropriate for principally 
operational documents. There have been a number of 
suggestions/volunteers that NANOG might provide a document function, 
but it seems reasonably clear that the Merit people don't have room 
on their plate for this.  Whether Euro-IX, RIPE, or RIPE-NCC are 
possible sponsors for operational documents, or if there needs to be 
a specific group spun off for this, is another question. Obviously, 
IDR alone can't answer that question, but I mention it here because 
many relevant people see this discussion.

>
>>  > The
>>  > fact that they did not get a wide vendor support is bug not a feature.
>>
>>  Please note that RFC 1863 doesn't so much define a route server as
>>  it does a route-reflector style mechanism for route servers.  This is
>>  the bit that didn't get wide deployment.
>
>Am sure IXPs would love to have better ways to implement route 
>servers. If there are any IXPs
>here, they could probably second me!
>
>>  IMO, 1863 is a bit more complicated than what we would want these
>>  days given modern route reflection techniques.  If a need exists to
>>  allow multiple views to be passed around in BGP there are better ways
>>  to go about it.
>
>Like how?
>
>Rasputin.
>
>
>______________________________________________________________________
>Post your free ad now! http://personals.yahoo.ca
>
>_______________________________________________
>Idr mailing list
>Idr@ietf.org
>https://www1.ietf.org/mailman/listinfo/idr

_______________________________________________
Idr mailing list
Idr@ietf.org
https://www1.ietf.org/mailman/listinfo/idr