May 1998 - Test - lists.nanog.org

Re: Dijkstra Simulator
by Daniel O. Awduche 12 Jun '98

12 Jun '98

The WANDL network planning and analysis software can be used for this purpose (and much more). Alternatively, you can trivially write an all "pairs shortest path" program, using, for example, the Floyd-Warshall dynamic programming algorithm (See "Introduction to Algorithms," by Corman, Leiserson, and Rivest). The same result can be gotten by running Dijkstra's algorithm |V| times (once for each node), where |V| is the cardinality of the set of nodes in your network graph. /Dan. Alan Hannan said: > > Could someone provide a reference for a Dijkstra simulator? > > Ideally this would take three sets of inputs (links, nodes, > metrics) and produce edge-pair traffic flows. > > Any suggestions appreciated, especially for GPL stuff that has > code. > > -a

3 2

flat vs non-flat charging
by Jerry Scharf 08 Jun '98

08 Jun '98

I think the idea of distance charging is going away in many cases. With WDM, the cost of the WDM and SONET eqiupment on the ends of a fully populated 32 channel per fiber, 144 strand pull vastly outweight end-to-end fiber costs of anything pulled through the ground. When you add routers and the like on top of that, the distance issue really goes away and it becomes on of network topology hops. Can anyone with figures for new intercontinental pulls say whether this is true there as well (project oxygen marketing claims this, but...)? Using archaic telephone pricing models to argue cost of providing bulk IP services is just not right. jerry

5 5

Re: IP over SONET considered harmful?
by avg＠pluris.com 07 Jun '98

07 Jun '98

> Subject: IP over SONET considered harmful? > > I'd hoped that MPLS would solve this problem, but from reviewing > the drafts I believe that the LSRs _WILL_ decrement the TTL. I think this is a non-issue, considering that with the next generation of IP-only boxes ATM will become hopelessly useless. Better yet, the real big boxes are removing any need to do clustering, so the average number of IP hops for an NSP is going to be like 3 or 4. --vadim

6 5

Re: different thinking on exchanging traffic
by Tim Salo 03 Jun '98

03 Jun '98

> From: "Sean M. Doran" <smd(a)clock.org> > Subject: Re: different thinking on exchanging traffic > Date: Fri, 22 May 1998 17:38:12 -0700 > [...] > Meanwhile, you might find the Sprint response to the NSFNET > NAP solicitation ("National Virtual NAP") hanging around somewhere > online. You should read it and imagine why it didn't fly. I believe that all four of the winning NSFNET NAP submissions proposed nationwide "NAPs". I believe that the reason they didn't happen is that the NSF asked for and assumed it would get four geographically-focused solutions. I suspect that the notion of awarding four NAPs, all of which covered all of the country, provided the NSF a certain amount of heartburn. I believe that the nationwide NAP concept died, (or was killed), at the time for administrative, not technical, reasons. But, this is all speculation on my part... I have two conflicting notions about the the interesting possibilities offered by nationwide layer-two services: o Layer-two services with distance-insensitive pricing, such as ATM, create some interesting opportunities. If it doesn't cost any more to get across the country than to get across town, why should I build a local NAP rather than a nationwide NAP? (Unless, of course, I am a RBOC and am administratively constrained from offering inter-LATA service.) (I am also ignoring a comparison of a NAP-in-a-closet/POP/parking ramp versus a NAP-in-a-metropolitan-area; this is e-mail to nanog, not a paper for Sigcomm.) Perhaps more relevant today, why should I build a regional Gigapop, _if_ my ATM pricing is truly distance-insensitive? (There might be an answer to the last question, I really don't know. But, I keep asking.) In other words, if pricing is distance-insensitive, why do I need local exchanges? o Distance matters. It is easy to configure an IP network over a large layer-two service that bounces packets around the country, (because IP routing protocols generally think in terms of hop count, not [physical] distance). It would be nice if routing protocols thought about [physical] distance, rather than require the network designer to be responsible for designing the network such that considerations of physical distance were implicit in the network design. Of course, in the good old days before distance-insensitive-priced services, this wasn't such an issue. (I don't know whether there are very many copies of Sprint's "National Virtual NAP" proposals hanging around. I had to nearly sign my life away to get a copy at the time...) -tjs

8 13

RE: Why does Sprint have address filters again?
by Jamie Scheinblum 03 Jun '98

03 Jun '98

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 make it "more fair" to the smaller networks looking to multi-home (See Karl's proposal on IP allocation). -jamie(a)networked.org -----Original Message----- From: Michael Dillon [mailto:michael@memra.com] Sent: Saturday, May 30, 1998 12:02 AM To: nanog(a)merit.edu Cc: arin-council(a)arin.net Subject: Re: Why does Sprint have address filters again? On Fri, 29 May 1998, Karl Denninger wrote: > Now, let's look at the parallels: > > 1. Both are required to "do business" in a given sector (ie: announce > routes, sell to the Erate customer base) > > 2. Both are simple *technical* providers (assignment of a number, with > the important being that it is unique in both cases). > > 3. One is free to the ISP. > > 4. The other costs $500.00 5. One is financed by the government out of your taxes and is merely an accounting formality much like a customer ID number. The other is funded by a corporation that has no government funding and must support itself not unlike most businesses and the number is a critical infrastructure identifier something like an NPA-NXX. > What is going on here? ASNs didn't used to cost money until ARIN got its > claws into them. ASNs have always cost money to issue. It's just that in the past it was funded out of taxes funnelled through the NSF to a subcontractor and hidden somewhere in NSI's budget. Those days are gone, thank God. -- Michael Dillon - Internet & ISP Consulting Memra Communications Inc. - E-mail: michael(a)memra.com http://www.memra.com - *check out the new name & new website*

7 13

Re: The Great Exchange
by shields＠crosslink.net 02 Jun '98

02 Jun '98

In article <19980528101417.10618(a)scfn.thpl.lib.fl.us>, "Jay R. Ashworth" <jra(a)scfn.thpl.lib.fl.us> wrote: > On Thu, May 28, 1998 at 03:19:26AM +0000, Michael Shields wrote: [...] > > Probably it will never make market sense to have distance-sensitive > > traffic pricing for "low-speed" users, where the cost of providing the > > service is mostly the cost of tech support, billing, dialin or xDSL > > aggregation, &c., and bandwidth is a small proportion of the cost. > > With current trends, will the actual bandwidth _ever_ be more than a > small fraction of the cost? Despite preductions, very few resources have ever actually become "too cheap to meter". Can you back up your comment about metering requiring three times the current CPU capacity of routers? -- Shields, CrossLink.

10 16

/24 in the RIPE netblocks
by Mikael Abrahamsson 02 Jun '98

02 Jun '98

Ripe delegates the 193/8, 194/8 and 195/8 blocks to european customers and some of these are /24 called "provider independant" nets and therefore arent connected to an ISP but rather directly to the customer and thus by definitition cannot be aggregated. The original idea was to have these so people wouldnt have to renumber when changing ISPs. I guess a lot of you consider this "a bad thing" (tm) but it's the way it is... My question is how many in the US filter out these /24 announcements from these blocks? And if you filter, would you consider making an exception for /24 in 193,194,195/8? ----- Mikael Abrahamsson email: swmike(a)swm.pp.se

3 2

PROPOSAL #2 - reform of the fee schedule (fwd)
by Karl Denninger 01 Jun '98

01 Jun '98

I am expanding the discussion list to NANOG and the ARIN Members mailing lists. Please post comments PUBLICALLY, or if you feel you cannot do so without "cloaking" yourself, through someone else (who you trust to strip the attribution). Towards a more fair fee proposal: I would like to ask the AC to direct Kim Hubbard to produce a list of the number of allocations in each of the fee categories currently listed, along with the documentation required to process each. >From this I believe we will find: 1) *TO BE FAIR*, one must document all allocations and use in all existing space at least bi-annually (once every two years) when coming back for more space. 2) A smaller allocation history naturally has less to produce. Therefore, I argue that the fees should be *proportionate* to the existing space. The smallest allocation, a /19, is 8192 addresses. Let's say that we set the "desired entry price" at $2,000. Therefore, the "cost per address allocated, including the current request", is set at, say, $0.25. Now Mr. Big Provider comes in and wants space. They have, let's say, the equivalent of an /8. They pay ~$4M under this formula. Obviously, we have just set the cost too high. Ok. ARIN has an operating budget of $2M. So we cut the base cost by 90%. Therefore, the /19 requestor pays $200 (!), or 0.025 cents per address IN USE, including the NEW requested allocation. Mr. Big Provider pays ~$400,000 Annually. The first allocation you request in a calendar year, you get assessed for ALL space you have, including the request you put in for. So if you have a /16, and ask for another /16, you get assessed for 65536 * 2, or 128k addresses. This works out to a fee of $3276.80. Note that if you ask for less, the fee is lower. There's your resistor on people asking for more than they need - it costs money (regardless of whether or not you get approved!) This is *FAIR*. We're now charging for the amount of verification work to be performed and EVERY allocation is treated equally. Now the reality is that this will put us WAY over budget. This is GOOD. What we need to do then is cut the base membership fee to something that ordinary people can afford - $50/year. The stakeholders should have a say in this, and the stakeholders, folks, are the average users. Would someone please tell me why this isn't a more proper fee strategy? -- -- Karl Denninger (karl(a)MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin http://www.mcs.net/ | T1's from $600 monthly / All Lines K56Flex/DOV | NEW! Corporate ISDN Prices dropped by up to 50%! Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost

4 4

USAC and SPIN
by Jim Fleming 01 Jun '98

01 Jun '98

@@@ http://www.neca.org/usnewrel.htm Whippany, NJ -- May 21, 1998 - The Universal Service Administrative Company (USAC) is pleased to announce the appointment of former Wisconsin Public Service Commission (PSC) Chairman Cheryl L. Parrino as USAC's first Chief Executive Officer (CEO), effective June 1, 1998." @@@@@@@@@@@@@ The U.S. Government provides ISPs with SPIN numbers, FREE of charge. Service Provider Identification Number (SPIN) @@@@ http://www.neca.org/funds/usacspin.htm "For companies which have not filed a FCC Form 457, you may contact USAC at (888) 641-8722 to obtain a SPIN assignment. @@@@@@@@@@@@@@@@@@@@@@@ Jim Fleming Unir Corporation - An H.323 GateKeeper for the IPv8 Network http://www.unir.net 0:196 .MALL

2 1

RE: Why does Sprint have address filters again?
by Jamie Scheinblum 01 Jun '98

01 Jun '98

The last CIDR report I have in my in-box states there are 3606 AS numbers in the routing table. So, my first question is at what point does ARIN reclaim AS numbers? (comparison: 2680 last September and 3075 at the beginning of the year) That's a lot of money Arin rakes in on un-used AS numbers. I understand that ARIN probably doesn't want to get in the middle of routing policies, but by limiting when you can receive an AS, I don't understand why they couldn't limit when you lose it. Second Q: How many AS numbers are available in total? Best regards, Jamie Scheinblum - FASTNET(tm) / You Tools Corporation jamie(a)fast.net (888)321-FAST(3278) http://www.fast.net FASTNET - Business and Personal Internet Solutions The views stated above are mine and do not reflect those of my employer. -----Original Message----- From: Karl Denninger [mailto:karl@mcs.net] Sent: Sunday, May 31, 1998 2:35 PM To: Patrick W. Gilmore Cc: nanog(a)merit.edu Subject: Re: Why does Sprint have address filters again? On Sun, May 31, 1998 at 12:30:46AM -0700, Patrick W. Gilmore wrote: > At 01:27 AM 5/30/98 -0500, Karl Denninger wrote: > >This does make sense - a lot of sense. > > >Karl Denninger (karl(a)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. > > 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. Uh, hold on a second.... I didn't say to make the first ASN "unreasonably" expensive (and I do believe $500 is unreasonable). However, with a REASONABLE first ASN fee (ie: $50 or thereabouts) bundling THAT with a /19 when you get your first PI allocation is even more reasonable. After all, the justification for the IP space encompasses that for the ASN, so the work has already been done, and the additional effort at that point should be literally a few keystrokes. My proposals to fix the issue with regards to getting a /19 if you're multihomed are also out there; has NANOG seen them? -- -- Karl Denninger (karl(a)MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin http://www.mcs.net/ | T1's from $600 monthly / All Lines K56Flex/DOV | NEW! Corporate ISDN Prices dropped by up to 50%! Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost

3 3