July 2000 - Test - lists.nanog.org

What's a termination
by Sean Donelan 18 Jul '00

18 Jul '00

On Mon, 17 July 2000, "Patrick W. Gilmore" wrote: > The telco world is completely different. *EVERY* call has an "origination" > and a "termination", and *EVERY* termination is *PAID*. Every "telco" termination is paid. However the telco's like to define some terminations as other than a telco, so the telco doesn't pay to terminate those calls. For example, in the USA calls to and from cellular carriers are charged access fees by the wireline carrier. Telco's first wanted to charge access fees for calls to ISPs, now they want a "Free-Ride"(tm) to terminate calls without reciprocal compensation to places they don't want to pay. I think a big difference is in the telco world, telco's collect more money for more call minutes they complete. If the call isn't completed, they don't get money. So telco's have an incentive to complete (and bill) as many call minutes as possible. Much of the US Internet is still flat-rate, so ISPs collect the same amount of money for most levels of service good or bad. ISPs have an incentive to deliver the least quality of service they can before losing customers. Even if an ISP has a multi-day outage, they bill for the service unless customers get so irate they may receive a token refund. Even ISPs with SLAs tend to make the refund procedure so cumbersome, they know its like giving out coupons instead of lowering the price. Only a small number of customers will ever get through the process and collect.

1 0

Telco Network Operations Forum
by Sean Donelan 18 Jul '00

18 Jul '00

On Tue, 18 July 2000, Robert Cannon wrote: > I am curious how often you think that ATT telephone long distance would hand > traffic off to MCI telephone long distance. My impression of telephone long > distance is that it is largely an end to end service (one of the great > differences from the Internet). That as far as long distance goes, there is > not a great deal traffic hand off (one exception is times of network trouble > where carriers have agreements to hand off traffic to maintain network > reliability). The urban legend has been MCI was AT&T's third largest customer (after General Motors and US Government). However as far as I know, there is no publicly available source of which telco uses what. Telco's swap a lot of things to reach different places. There are 1800 local telephone companies, but IXCs like Sprint or Worldcom only have a few hundred POPs around the country. So frequently an ILEC (usually the former RBOC) provides tandem service between an IXC and another LEC. Or an IXC may backhaul the call across another IXC's circuits. Look back to many of the "cream-skimming" complaints. Add in complications like Illuminet's SS7 service between multiple carriers, and a simple problem can involve a lot of different carriers. As Patrick mentioned, international calling is even more likely to involve "hidden" carriers. An inter-exchange carrier may not have its own international agreements with every carrier in every country. So if you are in PacBell land with Qwest as your long distance provider and call some place like Cuba it may pass through several other carriers or countries on its way to Cuba. And for odd requests, such as a person-to-person call to a sub-sahara country, the IXC will punt the call and tell you to call AT&T for help. I've heard of one international circuit between New York and Stockholm involving 13 different carriers. And no, the telephone carriers don't speak to each other any better. ATIS has an operations group which maintains an extremely thick manual how telco network management and control centers are supposed to work together. If you read the minutes from their meetings, it sounds remarkably like polite NANOG. The problems are the same, the words just have more syllables. Read Cliff Stoll's book The Cuckoo's Egg for a fun description of how many telephone companies may be involved in a call as he tried to trace his intruder. The National Communication System compiled a database covering the physical connectivity of the three major IXCs and dominant ILECs under an non-disclosure agreement for "national security/emergency preparedness" purposes. But even that is incomplete. And woefully inadequate for trying to predict what would happen to the Internet because most Internet circuits seem to run on the new carriers with cheap bandwidth instead of the three major IXCs. For example, UUNET uses some non-MCI/Worldcom circuits, as you can often see when Worldcom has a fiber cut isolating the Worldcom POP, but UUNET is unaffected.

1 0

Re: Telco NOC vs. Internet NOCs (WAS: Wanted: Clueful Individual @ TeleGlobe.net
by Robert Cannon 18 Jul '00

18 Jul '00

I am curious how often you think that ATT telephone long distance would hand traffic off to MCI telephone long distance. My impression of telephone long distance is that it is largely an end to end service (one of the great differences from the Internet). That as far as long distance goes, there is not a great deal traffic hand off (one exception is times of network trouble where carriers have agreements to hand off traffic to maintain network reliability). Also sprach Patrick W. Gilmore >The telco world is completely different. *EVERY* call has an >"origination" and a "termination", and *EVERY* termination is *PAID*. Example 1: User A goes off hook to call user B who is in another state. User A's local carrier is Bell South. Long distance is Qwest. User's B local carrier is PAcBell. User A pays a bill to Bell South and Qwest. Qwest pays an origination and termination access fee to Bell South and PacBell. User B pays PacBell. So there are flat rate charges on local networks, and metered origination and termination fees on the long distance network. Example 2: User A goes online to "webphone" and using A's webbrowswer calls User B. User's A Net access is not over the Public Telephone Network. "Webphone" is a free service in affiliation with GTE. User B subscribes to PacBell. User A pays a flat rate bill to its Net access provider; user A pays not long distance fee. GTE could act as a CLEC or a long distance company. If it acts as a long distance company IXC then GTE pays a termination access charge but NO origination access charge! User B pays a bill to pacbell. Example 3: User A calls User B 100% over the net and there are no telco charges. Example 4: User A calls User B. Both are on the Sprint Wireless network. Both pay flat rate charges to Sprint for 300 minutes of time. Since the entire call is on network, there are no access charges and reciprocal compensation. If at one end there are two local networks passing traffic, then the local networks must pay a metered reciprocal compensation charge. There are also universal service fees involved. Point: You cannot say "every" call has certain fees. The telephone network gets more diverse by the moment. If there is a trouble with the service in the pacbell network, what exactly do you want bell south to do about it? I guess I an not sure I understand. Finally, dont hestitate to drop essentially an email to the FCC letting them know trouble that you are having with your service. http://www.fcc.gov/eb/tcd/complaints.html Thanks Robert Cannon Internet Telecom Project www.cybertelecom.org ------Original Message------ From: Jeff Mcadams <jeffm(a)iglou.com> To: nanog(a)merit.edu Sent: July 18, 2000 12:27:11 PM GMT Subject: Re: Telco NOC vs. Internet NOCs (WAS: Wanted: Clueful Individual @ TeleGlobe.net Also sprach Patrick W. Gilmore >The telco world is completely different. *EVERY* call has an >"origination" and a "termination", and *EVERY* termination is *PAID*. ______________________________________________ FREE Personalized Email at Mail.com Sign up at http://www.mail.com/?sr=signup

2 1

ARIN to Allocate from 65.0.0.0/8 and 66.0.0.0/8
by Richard Jimmerson 18 Jul '00

18 Jul '00

In the near future, ARIN will begin making allocations from 65.0.0.0/8 and 66.0.0.0/8. This will include allocations of /20 and shorter prefixes, according to ARIN's minimum allocation policy. For informational purposes, a list of ARIN's currently administered blocks can be found at http://www.arin.net/regserv/IPStats.html Regards, American Registry for Internet Numbers (ARIN)

1 0

Telco NOC vs. Internet NOCs (WAS: Wanted: Clueful Individual @ TeleGlobe.net)
by Patrick W. Gilmore 18 Jul '00

18 Jul '00

At 10:14 AM 7/17/00 -0700, Karyn Ulriksen wrote: > >Out of curiousity, how does the telco world handle this? When some one >signs up for MCI longdistance, I assume that on some calls they will cross >off the MCI network (and potentially though several others) that will be >piled mile high in static. The customers will complain about their >connection to their respective local and long distance companies. But I >believe it isn't an option to contact the networks that they may transit >mid-stream. However, some QC must be in place to achieve some level of >consistancy to get that expected "pin drop" quality we've apparently come to >take for granted... The telco world is completely different. *EVERY* call has an "origination" and a "termination", and *EVERY* termination is *PAID*. So when it goes from you (user) to AT&T to MCI to Telstra to friend (user), you pay AT&T, AT&T pays MCI, and MCI pays Telstra. (I made that sequence up. I doubt a call would be routed like that, but you get the point.) So, if there is a problem inside MCI, AT&T will hound them about it. Even though no "user" is paying MCI, AT&T *is* paying them, and has every right to get upset if quality is not up to par. Plus, the user has no idea of routing in the phone system. There is no traceroute, there are no in-addrs, etc., etc. As a user, you call AT&T, and that is all you know. But that is okay, because AT&T takes full responsibility for end-to-end connectivity. If the call does not get through, they may say "MCI is having a problem", but they will work on it for you. Some may say this is why the telcos work better than the Internet. Personally, I think it is kinda silly. The way the system is laid out, it costs a lot more to bill a call than connect a call. It takes a lot more to track the minutes of a call and all the various tariffs, interconnections, time-of-day billing, billing cycles, interest / late fees, mail out a bill every month, etc., etc., etc., than it takes for a few switches to route a call over fiber which costs nearly nothing anyway. Remember, typical pricing for trans-continental US fiber, when buying dark or very large IRUs, is a few places to the right of the decimal point per month per DS0-mile - or *at most* $3/month for a cross country DS0, and for most big carriers it is orders of magnitude less. And the telcos tend to get a wee bit more than $3/month revenue off that DS0. ;) Of course, there is cap-ex for the switches & fiber lighting equipment - which is a huge expense. Plus people to operate it and stuff like that. But the point is, it costs more to bill it than to actually *do* it. I do not think we want to see the Internet go that way. Do you? I am not saying this is a valid way to run the Internet, just that it is not how I would like to see the Internet run. Perhaps my view / opinions are silly or impractical, but that is how I feel. What do all of you think? >Karyn TTFN, patrick

2 1

RE: Telco NOC vs. Internet NOCs (WAS: Wanted: Clueful Individual @ TeleGlobe.net)
by rdobbins＠netmore.net 18 Jul '00

18 Jul '00

You're certainly correct in your explanation of tolls, tariffs, etc.; however, do bear in mind that the telcos are heavily regulated, that they have quality-of-service and 'uptime', if you will, standards they must meet from the various agencies who impinge upon them (both Federal and local), and that they also must get both price increases -and- decreases approved at the various affected levels, as well. All these things are part of what contribute to the Soviet-style mentality found within most telcos. They're also part of the reason that - with very few exceptions - telephone service from the user perspective works far more reliably than Internet connectivity. I personally would never want the Net to go the way of the telcos, after having worked for one. Conversely, there -are- some lessons about quality-of-service which the ISP world can learn from them. My greatest fear in this regard is that as Internet connectivity becomes something that everyone expects and needs, just like their telephone service (or even becomes their telephone service), that the regulators will step in, unless we all can get our acts together enough to ensure that they've nothing to worry about. I'm not naive enough to think that they won't try to regulate our operations, anyway, but if we do our best to get our own house in order, they'll have less of a change of actually succeeding. ----------------------------------------------------------- Roland Dobbins <rdobbins(a)netmore.net> // 818.535.5024 voice -----Original Message----- From: Patrick W. Gilmore [mailto:patrick@ianai.net] Sent: Monday, July 17, 2000 9:07 PM To: nanog(a)merit.edu Subject: Telco NOC vs. Internet NOCs (WAS: Wanted: Clueful Individual @ TeleGlobe.net) At 10:14 AM 7/17/00 -0700, Karyn Ulriksen wrote: > >Out of curiousity, how does the telco world handle this? When some one >signs up for MCI longdistance, I assume that on some calls they will cross >off the MCI network (and potentially though several others) that will be >piled mile high in static. The customers will complain about their >connection to their respective local and long distance companies. But I >believe it isn't an option to contact the networks that they may transit >mid-stream. However, some QC must be in place to achieve some level of >consistancy to get that expected "pin drop" quality we've apparently come to >take for granted... The telco world is completely different. *EVERY* call has an "origination" and a "termination", and *EVERY* termination is *PAID*. So when it goes from you (user) to AT&T to MCI to Telstra to friend (user), you pay AT&T, AT&T pays MCI, and MCI pays Telstra. (I made that sequence up. I doubt a call would be routed like that, but you get the point.) So, if there is a problem inside MCI, AT&T will hound them about it. Even though no "user" is paying MCI, AT&T *is* paying them, and has every right to get upset if quality is not up to par. Plus, the user has no idea of routing in the phone system. There is no traceroute, there are no in-addrs, etc., etc. As a user, you call AT&T, and that is all you know. But that is okay, because AT&T takes full responsibility for end-to-end connectivity. If the call does not get through, they may say "MCI is having a problem", but they will work on it for you. Some may say this is why the telcos work better than the Internet. Personally, I think it is kinda silly. The way the system is laid out, it costs a lot more to bill a call than connect a call. It takes a lot more to track the minutes of a call and all the various tariffs, interconnections, time-of-day billing, billing cycles, interest / late fees, mail out a bill every month, etc., etc., etc., than it takes for a few switches to route a call over fiber which costs nearly nothing anyway. Remember, typical pricing for trans-continental US fiber, when buying dark or very large IRUs, is a few places to the right of the decimal point per month per DS0-mile - or *at most* $3/month for a cross country DS0, and for most big carriers it is orders of magnitude less. And the telcos tend to get a wee bit more than $3/month revenue off that DS0. ;) Of course, there is cap-ex for the switches & fiber lighting equipment - which is a huge expense. Plus people to operate it and stuff like that. But the point is, it costs more to bill it than to actually *do* it. I do not think we want to see the Internet go that way. Do you? I am not saying this is a valid way to run the Internet, just that it is not how I would like to see the Internet run. Perhaps my view / opinions are silly or impractical, but that is how I feel. What do all of you think? >Karyn TTFN, patrick

1 0

Wanted: Clueful Individual @ TeleGlobe.net
by Derek J. Balling 17 Jul '00

17 Jul '00

OK, I've tried the normal methods... phone, email... no response... Sample snip from phone call: Me: I wanted to report some network issues on your network. You're showing some pretty major packet loss between <ip> and <ip> on your network. TG: What's your circuit ID? Me: I don't have a Teleglobe ckt, I'm just some average schmoe who has noticed that you're dropping packets on the ground at a VERY alarming rate. TG: I can't enter a trouble ticket for you unless you're a teleglobe customer. Me: Don't enter it for me, enter it for yourself, its your network that's got the problem. TG: I can't do that. Me: So even though your network is sucking wind right now and hemorrhaging packets, you won't start up a trouble ticket simply because I'm not PAYING you to take a trouble ticket? TG: Ummmm, correct. Me: If you were a Teleglobe customer, and knew that someone had reported this problem but you ignored the request because the person telling you about it wasn't a customer, would you be pissed? TG: Ummmm, probably. Me: So put your fingers to the keys, and do the RIGHT thing, and open up a ticket on it. TG: I can't. Me: OK, let me speak to your supervisor. TG: I can't. I'm the supervisor tonight. Me: Is there ANYONE more clueful above you in the chain of command that I could hope to speak to in order to try and help you guys from putting yourselves out of business? TG: Ummmm, no not right now. Me: When would there be more people in. TG: Monday. Me: Bah, by then, you'll be out of business. Cheers. FYI, their problem is here: Hostname %Loss Rcv Snt Last Best Avg Worst 5. atm8-0.core1.paix.teleglobe.net 0% 32 32 78 78 141 480 6. if-4-0-0.bb5.NewYork.Teleglobe.net 55% 14 32 172 161 195 437

8 9

to learn about routing.
by Vaul Pixie 17 Jul '00

17 Jul '00

For those interested in learning more about basic routing, I would like to direct your attention to the follow book, "Internet Routing Architectures, Second Edition, Volume I", this fine text can be ordered from many bookstores. Below is a purchase order for it from Amazon, filled out to one, Hansen Chan. I'm sure you will be just as pleased with it as he was. Vaul Pixie vaul(a)wayabove.net ---------- Forwarded message ---------- Date: Sat, 17 Jun 2000 03:39:36 -0400 From: orders(a)amazon.com Subject: Your Order with Amazon.com (#xxx-xxxxxxx-xxxxxxx) Thank you for ordering from Amazon.com. Your purchase information appears below. <snip> Order #1: (Order #103-4820841-6376601) Hansen Chan Alcatel Network Systems 1225 N. Alma Road, MS 401-102 Richardson, TX 75081-2206 United States Shipping method: Standard Shipping Shipping Preference: Ship this when complete 1 "Internet Routing Architectures, Second Edition, Volume I" Bassam Halabi; Hardcover; @ $55.00 each Usually available in 24 hours ---------------------------------------------- ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com

5 4

***Announcement: Merit to convert to 'PGPKEY' authentication (RFC 2726)
by gerald＠merit.edu 17 Jul '00

17 Jul '00

On 7/30/00 between 5 AM - 8 AM EDT Merit will convert from 'PGP-FROM' to 'PGPKEY' authentication as described in RFC 2726. All maintainer object 'PGP-FROM' instances will be converted to a 'PGPKEY' instance and a 'key-cert' object will be created for each unique 'PGPKEY' instance. After the conversion all public keys used for signing DB submissions will be visible in the registry as 'key-cert' objects. Registering public keys allows users to verify a maintainer's identity and to submit, update and maintain their own public keys like any other DB object. Normal PGP signing of DB submissions will continue to work exactly as before the conversion. The conversion should be seamless to our DB user community and no action is required. Merit will send email to all users who will be affected by the conversion. The notice will give users a chance to remove their PGP references if (for whatever reason) they do not wish to take part in the new PGP authentication scheme. The new 'PGPKEY' scheme makes it easier for users to integrate PGP signing into their DB update process. It is our wish that into the future all registry users will PGP sign their DB submissions. Please see www.radb.net/docs/keyfaq.html for additional information on the conversion process and tutorial information regarding the use of 'PGPKEY' authentication. Please consult RFC 2726 for the authoritative specification. IRRd team --Gerald Winters

1 0

RADB and building router filters
by Jim Mercer 17 Jul '00

17 Jul '00

i'm running into difficulties getting my routes announced via networks that use the RADB to build their filters. i'm being told that some of my routes are being ignored because there are supercede'ing entries at whois.radb.net. for example, "whois -h whois.radb.net 198.96.117.0" contains two entries: route: 198.96.117.0/24 descr: NETBLK-TORMAN origin: AS813 mnt-by: MAINT-AS813 changed: support(a)uunet.ca 19950920 source: RADB route: 198.96.117.0/24 descr: Reptilian Research admin-c: JM7718-ARIN tech-c: JM7718-CSAS origin: AS843 remarks: in use by Reptilian Research remarks: parent is ARIN notify: jim(a)reptiles.org mnt-by: MAINT-AS843 changed: jim(a)reptiles.org 20000527 source: CSAS should i be requesting that the entry from UUNET be removed? -- [ Jim Mercer jim(a)reptiles.org +1 416 410-5633 ] [ Reptilian Research -- Longer Life through Colder Blood ] [ Don't be fooled by cheap Finnish imitations; BSD is the One True Code. ]

3 4