Why does Sprint have address filters again?

Patrick W. Gilmore patrick at priori.net
Sun May 31 07:30:46 UTC 1998

At 01:27 AM 5/30/98 -0500, Karl Denninger wrote:
>This does make sense - a lot of sense.

>Karl Denninger (karl at MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin

>On Sat, May 30, 1998 at 12:26:31AM -0400, Jamie Scheinblum wrote:
>> Suggestion:
>> The initial ASN should be bundled with a /19 to create a "multi-home"
>> package.  Unbundled ASNs whould be unreasonably high to cover the
>> administration of the initial ASNs of the world, and the cost associated
>> with a /19.
>> In reality it seems you need both, a /19 to make it past the route filters,
>> and an ASN.  This also save on the ARIN support side, since the ARIN
>> employee tasked with making the call to verify the customer does in fact
>> have 2 T-1s (or has 2 ISPs vouch he will have at least a T1 with each), can
>> also verify they will accept the routes for the ASN.
>> Seems like this would cut the administraion on ARIN's behalf a bit, and
>> it "more fair" to the smaller networks looking to multi-home (See Karl's
>> proposal on IP allocation).
>> -jamie at networked.org

No, it does not make a lot of sense.  In fact, IMHO, this is a very, *very*
bad idea.

Remember what we are trying to conserve here - not just ASNs, but IP Space
too.  Making unbundled ASNs "unreasonably high" would kill all the people
who are being good little businesses and conserving IP space.

For instance, what do you say to the large corporation who wants the
redundancy of two connections, but uses NAT or DHCP so they only need a /23
or /22?  And what do you say to the web farm that uses HTTP 1.1 to
conserver IP space but who's customers demand redundancy or low latency to
several networks?

There are many, many more instance of perfectly legitimate uses for an ASN
with less than a /19.  As for the "filtering" issue, well, Sean and I have
had words over that before.  The large majority of my customers multi-home,
and less than a third of those have (or need) their own /19.  They are
aware than they'll only get Sprint & AGIS traffic from my network (because
I announce the whole CIDR), and that if my T1 drops the packet will go from
Sprint to Priori to their other upstream in a round-a-bout fashion.  But
they generally feel this is an acceptable penalty to pay for the
conservation of IP space.

Are you suggesting that we penalize these people even more for doing the
"right thing"?  The could drop all those conservation tactics and say they
"need" a /19, thereby depleting the v4 space that much sooner - which I'm
sure they'd do if you made the price "unreasonably high".

On the flip side, charging more (a lot more) for 2nd, 3rd, etc. ASNs sounds
fine to me.  Admittedly, I have not thought the whole thing through since
neither I nor any of my downstreams use multiple ASNs, so I might be
missing something.  Feel free to correct me.

It just seems to me that multiple ASNs *might* be necessary in some
instances - like when multiple countries are covered under one network.
But in the overwhelming majority of cases, it seems that things like BGP
confederations could be used internally for anything that required multiple
ASNs, while letting the rest of the 'Net see just one ASN.  Sprint, for
instance, has many, many ASNs.  Considering their (Draconian? ;) policy of
conservation wrt IP space, why do they suddenly need dozens of ASNs?

For the record, I do not think the $500 one-time fee is completely
ridiculous.  It might be lowered, but it's not nearly as outrageous as some
people are suggesting - IMHO.  And I really don't see how $30/year can be
considered usury by anyone.  Unless you would rather bill people on a "per
query" basis or something?  Say $0.01 per "whois"?  Or something equally
silly and ridiculously difficult to collect, bill, deal with delinquencies,
etc., etc.

Now the schedule of $5,000 per year just to have a block of space which is
capable of passing those filters we talked about....  Well, let's just say
that I'm not completely satisfied this is as reasonable.  Especially
considering it is an annually recurring charge, unlike the $500 ASN fee.


Patrick W. Gilmore                      voice: +1-650-482-2840
Director of Operations, CCIE #2983        fax: +1-650-482-2844
PRIORI NETWORKS, INC.                    http://www.priori.net
              "Tomorrow's Performance.... Today"

More information about the NANOG mailing list