Perhaps I owe some apologies for being harder on Sprint last night than I
should have been. Is there a routing arbiter list to which I should take
this discussion?? If so please let me know where and how to join and
I'll gladly oblige. In the meantime let me summarize what I am learning.
Jeff Hayward points out:
The NSF funded regionals, not NSPs. There is no requirement in
NSF93-52 that makes use of the RA mandatory. The only requirement
placed on the regional's choice of NSP is connection to all priority
NAPS.
Cook: But apparently NSF funded R&E regionals must ALSO make their
routes available to the RA?
For a friend at NSF writes privately to me that:
There is no requirement that anyone "use" the services of the RA. Specific
requirement is that:
Regional Networks and their selected NSPs must route and carry all
traffic originated at and/or destined for U.S. Research and Education
sites. Moreover, Regional Network Providers must make their routes for
all such (U.S.Research and Education) sites within their (respective)
regions available to the Routing Arbiter.
Then Jeff Barrows makes and interesting and useful point:
The last time I checked, the _biggest_ argument against using the RA was that
alot of the data is incorrect. I have also heard a lot about work that has
been done to clean up the RA. Is this still the biggest factor?
-Jeff
From Nanog archives jeff barrows cites sean doran as having said:
I would point out one more thing though, and that's that
at the Stockholm IETF I had a genial chat with a number
of folks from MERIT and the RA Team in general,
and suggested several ways that the RADB could
be made incrementally useful. I hope that some good
comes out of that conversation.
I'll use any tool that will make my job easier, and
help our operation and the Internet scale better.
At the moment, though, the RADB does the opposite,
and the RS has no value whatsoever.
Sean.
Now Bill Manning didn't NAME sprint yesterday in his complaint that I
have cited. But no one has told me that anyone else besides sprint was
intended for the criticism. Perhaps the question boils down to those
raised by Sean in the preceding paragraphs? What does need do be done to
the RADB and RS to give them value? There seems to be some strong
disagreement between the Routing Arbiter and Sprint. Why? What do they
see so differently? And if MCI, PSI, UUNET and ANS don't agree with
sean's criticisms why don't they?