- NANOG - lists.nanog.org

NANOG Host for October
by William B. Norton 01 Jun '97

01 Jun '97

Hi - Thanks to those that have volunteered for hosting NANOG meetings in the past. We now have an opening for the October '97 NANOG 11 slot. If you are interested in hosting this NANOG, please read the "Tips for Hosting a NANOG" (http://www.nanog.org/tips.html) document on the www.nanog.org web page and send me e-mail. Thanks! Bill -------------------------------------------------------------- William B. Norton <wbn(a)merit.edu> (313) 764-9430

1 0

Re: RFC 1918 addresses
by Joseph C. Pistritto 01 Jun '97

01 Jun '97

While this is technically correct, actually changing the addresses in ICMP payloads violates my first rule of debugging complex systems: Diagnostics should be a simple as possible and never be altered by any non-diagnostic system. (Which begs the question of should people be using these addresses locally at all. Which is kind of metaphysical except that it *does* preserve address space, which is a universal good. Some people do this for security reasons too, although it's at best only moderately effective security measure and could produce a false sense of security inside such an addressing realm.) I agree that ever having a source or destination IP that's RFC1918 outside the domain is a very bad thing. -jcp- ---------- > From: Paul A Vixie <paul(a)vix.com> > To: nanog(a)merit.edu > Subject: RFC 1918 addresses > Date: Saturday, May 31, 1997 4:38 PM > > I think that any exposure of these addresses outside their addressing realm > is a mistake. Using them for otherwise unnumbered serial links would be fine > if Cisco had an option to "use loopback address when sending ICMP's" but they > don't. Traceroute is sending increasing-TTL'd UDP datagrams, and if you are > seeing a 10.0.0.2 source address on an router's ICMP to you it means you would > get that as a normal ICMP too -- meaning not just one due to a diagnostic like > Traceroute. I think this is an error. > > Exposing an RFC 1918 private address in, say, a "Received:" header in e-mail > is less of a problem, though the spammers who do it are actually better able > to cover their origins, there's no way to prevent it and no normal damage > from doing it. > > No IP datagram with an RFC 1918 address in the protocol headers should be > allowed outside a private addressing realm. This means not the source > address, not the destination address, and not the encapsulated addresses > inside an ICMP payload.

2 1

Re: connectivity outside the US
by smd＠clock.org 01 Jun '97

01 Jun '97

"Miguel A.L. Paraz" <map(a)iphil.net> writes: > There can be two ways of looking at it, from the WWW/content point of > view: > > * "They're unlucky enough to be outside North America, but they want > American content anyway. So they have to suffer with poor > performance and high prices." > * "These folks outside America don't get to appreciate our content. > Their numbers are growing. Thus we have to invest in the delivery > infrastructure." Of course that leads to: * We have to worry about international copyright issues as overseas ISPs begin duplicating our content to save on IPL bandwidth * We should deploy duplicate servers in other parts of the world so that we don't have to deal with the problems of shortages of IPL bandwidth On the other hand I argue that the fundamentally interesting thing about the Internet is that traffic patterns shift substantially at little notice through the development and deployment of new technology. Moreover, in in Internet whose origins were "every client is also a server" and where, with proxying and the like, that is still essentially practical, a strict content:consumer mindset is a very short-sighted business approach. > But, the Independent Big Network people will still have to get > their own IPLs - which may end up with BT/MCI, Sprint/Global One, > or AT&T? What's the difference between becoming an IP customer > and just a leased line customer? Independent Big Network people are working on a simple risk statement: traditional bandwidth-owning telcos are too slow to keep up in the Internet marketplace, particularly when they're distracted by what various bits of deregulation are doing to their traditional income, creating an opportunity for the Independents either to thrive on their own or to be strong acquisition targets. Moreover, one might bet on partnerships with organizations competing with established telcos on other fronts which are becoming more like commodity products (most notably raw bandwidth). > How do we come up with a settlement model? Ultimately this depends on what everyone's real costs are. Alot of what is perceived as "cost" is due to misapplication of technology. A settlement model that encourages all parties to do the right thing to sustain the industry's growth is really what's needed. The problem is getting such a model agreed to and put into effect other than bilaterally among clever operators. Hahaha, good luck if there are American lawyers involved. > And more providers should be hauling in their own high-capacity pipes and > putting up their own overseas POPs instead of having the local folks > drag their lines to the US, and charge a lot for slow > connections. It's coming. The problem is that the people most likely able to do this are also the people earning half-circuit revenues from the large numbers of IPLs that would get aggregated into an IPL where both half-circuits would be paid by the provider entity. This is especially touchy when money for one of the half circuits is paid to another telco. In this way large capacity-owning telcos are in a bind about competing with a lucrative revenue stream (IPLs) whenever they do this and also when they compete against non-telco Independent Big Networks who might buy IPLs at market value. To make up for lost revenue (and maybe real cost) from the internally-deployed IPL, lots of traffic would have to be aggregated. This kind of massive overselling of capacity is still somewhat alien to the managed services mindset and even if it didn't it doesn't make it all that much easier for them to compete on "quality of service" profitably just now without some magic extra-high-quality pricing bullet. Sean.

1 0

RE: connectivity outside the US
by Sean M. Doran 01 Jun '97

01 Jun '97

Barry has an excellent point, and I'm glad that someone with his, um, background is helping undersea cable operators and Internet operators understand each other better. The not very extensive history of private undersea ventures has revolved around "cream-skimming" -- taking high-revenue high-capacity traffic from consortium-built cables -- and bandwidth appreciation. Essentially the private cable operators want to walk a line between having to depreciate their assets due to lack of revenue and having enough spare capacity for those times when consortium partners run short. That is, if you have a private cable, you don't want to sell all your capacity to end users because you are hoping that your competitors will run short on bandwidth, allowing you to jack up your prices. If and when you sell capacity you want it in long-term contracts, preferably for restoration of other paths so that if someone builds a competing cable or the accountants start demanding some return on investment you can do short-term "as-available" deals with government-funded research networks. Unfortunately this model does not tend to drive down prices substantially for most users, and particularly not for most users looking for an easy growth path (any kind of easily-tuned end-to-end TDM in theory should change this). Even turning unused capacity over for internal commercial purposes (and it is unclear whether UUNET would be an "internal purpose" given the joint ownership of the Gemini company) would be a huge risk as it involves tying capacity to a revenue stream, which means making hard decisions when someone wants to pay real money for that capacity when they run into a crunch. Terrestrial capacity politics have some similarities, btw, however the acquisition of rights-of-way in some places is so difficult that alot of the cross-leasing games revolve around trading capacity along various paths as much as playing the you're out of bandwidth pay us lots of money games (which AT&T does remarkably well, btw, in the U.S.). What becomes interesting is alot of new rights-of-way opening up and making it possible to offer dark fibre along them. This directly threatens the price escalation model as well as the managed-services pricing schemes bell-heads like. This is good for people building parts of the Internet. Sean.

1 0

RE: connectivity outside the US
by Barry Raveendran Greene 01 Jun '97

01 Jun '97

Hello Chris, > Worldcom and Cable & Wireless are in the process of laying a 20 Gbps > transatlantic sonet (DWDM) system which they claim doubles the current > transatlantic capacity. The southern route should be done by November > of this year with the first capacity available for the end of the year. > The northern route, which will close the ring, will be done in '98. Only 20 Gbps? KDD is putting a 100 Gbps around Japan. Due in 1998. ;-). Barry -- -- -- Barry Raveendran Greene | || || | Senior Consultant | || || | Consulting Engineering | |||| |||| | tel: +65 738-5535 ext 235 | ..:||||||:..:||||||:.. | e-mail: bgreene(a)cisco.com | c i s c o S y s t e m s |

1 0

RE: connectivity outside the US
by Barry Raveendran Greene 01 Jun '97

01 Jun '97

Hello Jesse, > Undersea capacity is expensive for 3 reasons: > 1) It's under the ocean > 2) It's under the ocean > 3) It's under the ocean Actually, based on first hand experience, the cost of installing a oceanic cable is _not_ the driving factor that makes oceanic bandwidth so expensive. The key factors involve the relationship between the partners of any one of the cable consortiums. Each of the partners are telcos. Telcos who compete with each other in the region do not have each other's interest at heart. Hence, when they set the prices for the cable capacity, the prices get jacked up. When ever one telco mis-forecasts and runs out of capacity on a cable, the other "partners" jack up their prices even further and sell it to the unfortunate "partner" who ran out of capacity. These pricing games between the telcos over oceanic bandwidth is one of the key factors for the high cost end customers experience. The problem with all of this is there is little information available that explains how oceanic bandwidth politics. This is why we've got a couple of blocked out sessions during APRICOT that covers oceanic and space segment (satellite) technology and politics. IMHO - Education is the best weapon to lowering the cost of oceanic bandwidth. Barry -- -- -- Barry Raveendran Greene | || || | Senior Consultant | || || | Consulting Engineering | |||| |||| | tel: +65 738-5535 ext 235 | ..:||||||:..:||||||:.. | e-mail: bgreene(a)cisco.com | c i s c o S y s t e m s |

1 0

RE: connectivity outside the US
by Barry Raveendran Greene 01 Jun '97

01 Jun '97

Hello Miguel, > Yes, the demand far outstrips the supply of bandwidth, we > are having that problem right now. Never mind that the we > can afford the > $60,000/month for a measly T1 to the US - > we are currently back-ordered as there is not enough available > channels across the Pacific. Actually, there is plenty of bandwidth coming out of the Philippines. The problem is that the telco you are working with did not do their bandwidth forecast correctly. Hence, they did not buy the enough allocations from the cable consortiums. This is why you are waiting. Your telco is waiting until the next cable consortium bidding cycle (June) to get more bandwidth allocated so they can sell it to you. The telco could get you your T1 now, but they would have to buy the capacity "out of cycle." This is expensive since they either have to buy the bandwidth off one of the other telcos in the region or at "out of cycle" rates from the cable consortium - both are very expensive. [lots deleted] > > or start coming to the same understanding: access to trans-oceanic > > capacity while it continues to be hiddeously expensive should be > > paid for at both ends, since both sides benefit. > > How do we come up with a settlement model? First you need the tools to measure flows. They did not exist several years ago. Today, with the work in RTFM (Real Time Flow Metering WG in IETF), Caida, Netflow, and other tools, the technical foundation required to create settlement agreements similar to the ones in the voice world are now possible. -- -- -- Barry Raveendran Greene | || || | Senior Consultant | || || | Consulting Engineering | |||| |||| | tel: +65 738-5535 ext 235 | ..:||||||:..:||||||:.. | e-mail: bgreene(a)cisco.com | c i s c o S y s t e m s |

1 0

RFC 1918 addresses
by Paul A Vixie 31 May '97

31 May '97

I think that any exposure of these addresses outside their addressing realm is a mistake. Using them for otherwise unnumbered serial links would be fine if Cisco had an option to "use loopback address when sending ICMP's" but they don't. Traceroute is sending increasing-TTL'd UDP datagrams, and if you are seeing a 10.0.0.2 source address on an router's ICMP to you it means you would get that as a normal ICMP too -- meaning not just one due to a diagnostic like Traceroute. I think this is an error. Exposing an RFC 1918 private address in, say, a "Received:" header in e-mail is less of a problem, though the spammers who do it are actually better able to cover their origins, there's no way to prevent it and no normal damage from doing it. No IP datagram with an RFC 1918 address in the protocol headers should be allowed outside a private addressing realm. This means not the source address, not the destination address, and not the encapsulated addresses inside an ICMP payload.

1 0

Re: connectivity outside the US
by Daniel McRobb 31 May '97

31 May '97

> smd(a)clock.org (Sean M. Doran) wrote: > > > Examining this a bit more closely, since undersea capacity is > > terribly expensive, when there is adequate capacity available > > to a large aggregate of sites people want to get to, there will > > be an obvious market for access to that capacity. > > Actually, i do not understand why undersea capacity is so > expensive. Cable is more expensive, yes; but the paths > are much straighter, and there's no need to purchase > rights of ways (except for shore-side strips). There's > no need to dig trenches -- you just drop the cable off the > boat. >From what I've read, it's quite difficult to lay undersea cable. There was a nice article in the Dec. 1996 issue of WIRED covering the FLAG (Fiberoptic Link Around the Globe) project, written by (of all people) Neal Stephenson. Online at: http://wwww.wired.com/wired/4.12/features/ffglass.html Daniel ~~~~~~

1 0

Re: 10.0.0
by Danny McPherson 31 May '97

31 May '97

> You can have an internal mesh made up of entireley rfc1918 address > space, and not leak these routes to the rest of the world, I've only > once caught MCI leaking stuff from a test lab, which was kinda annoying, > but not really anything bad, and a polite e-mail message to them got > an immediate fix of the problem. > i'd think most providers filter rfc1918 addresses both inbound and outbound at naps (mci does, i believe), although maybe not to customers... using reserved address space internally (as long is it remains internal), seems like a good idea to me - isn't that why it was reserved? -danny

2 1