[9890] in Commercialization & Privatization of the Internet
Re: I must be stupid, or asleep.
daemon@ATHENA.MIT.EDU (Karl Denninger)
Tue Jan 25 18:55:54 1994
From: karl@mcs.com (Karl Denninger)
To: nelson@crynwr.com (Russell Nelson)
Date: Tue, 25 Jan 1994 17:50:12 -0600 (CST)
Cc: com-priv@psi.com
In-Reply-To: <2d4592a0.crynwr@crynwr.com> from "Russell Nelson" at Jan 25, 94 04:43:59 pm
>
> I must be stupid, or asleep, or just plain not-paying-attention.
> Ittai says that the CIX-only non-NSF-AUP-signing crowd is shrinking,
> and he thinks that's the CIX's fault. (Actually, the CIX's fault is
> somewhere underneath their router in California).
>
> That makes perfect sense to me, as I understand today's
> state-of-the-art Internet routing. If I'm at a Research and/or
> Education (R&|E) site, I'm not doing business (by definition) and so the AUP
> is no problem. If anyone wants to do business with me, that's fine
> because they're supporting me, and that's okay.
>
> If I'm NOT at an R&|E site, and I want to talk to an R&|E site,
> that's okay because I'm supporting them. So there's no problem if I
> sign the AUP. If I'm NOT R&|E, and I want to talk to another
> business who isn't, then the AUP doesn't apply because our traffic
> isn't traversing the NSFNet. If the traffic does traverse the
> NSFNet, then it's not our problem, because we've paid for AUP-free
> Internet access.
>
> So the only people who would NOT sign the AUP are those people who
> are using the Internet as a replacement for a private network. That
> is to say, they don't care about general IP connectivity. They've
> just bought into a cheaper bit-pipe.
>
> It only stands to reason that the people in the latter group would
> shrink as they discover the advantages of Internet access. Would
> someone please reassure me that I'm not stupid, asleep, or just plain
> not-paying-attention?
You're making perfect sense.
In fact, that's exactly what is going on.
That someone is routed over the ANS backbone does NOT mean that is their
primary "network announcement preference". In fact, nothing of the kind is
implied by that fact. All it means is that they've signed up to be
AUP-complient when on the backbone, which most people are.
FEW people want to be CIX-only. Its an option I offer to people, but most
choose not to go that route.
MCSNet, in fact, routes more of our traffic via the CIX links than via the
T3 ANS lines. The fact that we have routes announced to the NSF core means
nothing as far as where our traffic actually <goes>.
This is a smokescreen. What you need to do is look at the number of
organizations which are <NSF ONLY>, and then those which are <CIX ONLY>.
THAT would be a valid comparison. To imply that there are some number of
NSF announced routes, and then some number of CIX announced routes, and
then to try to draw a comparison from that, is meaningless.
Both of those numbers are going to be <vanishingly> small.
See, the CIX is an association. MCSNet announces its routes under AS 3365,
which is ours. We source them to Sprint, on one of their AS numbers, and
they get redistributed from that point. Thus, we don't technically "source
our routes through the CIX", although the CIX picks them up <and uses them>.
That is because, unlike the NSFNet, there is no PRDB for the CIX.
The MERIT PRDB is a central <collection point> for NSFNet routes. Thus, it
has <ALL> the R&E routes in it (other than the CO+RE customers perhaps).
There is NO SUCH REGISTRY IN ONE PLACE for CIX routes.
Apples to apples please Ittai.
If you want to make this point, dump the BGP routing tables for the CIX
core and the ANS core (by a neutral party, please) and count the number of
networks which are resident in each. THEN you have a reasonable estimate
of the networks "reachable" by each path. You're going to get a <lot>
of overlap in that comparison, which is fine. All that means is that there
is more than one path to a particular point. This is <good>.
The number of networks "announced" by the CIX is a non-issue. Member's
networks are announced by the members, not the CIX itself.
--
Karl Denninger (karl@MCS.COM) | MCSNet - Full Internet Connectivity (shell,
Modem: [+1 312 248-0900] | PPP, SLIP and more) in Chicago and 'burbs.
Voice/FAX: [+1 312 248-8649] | Email "info@mcs.com". MCSNet is a CIX member.
