- NANOG - lists.nanog.org

Massive fiber cut ??
by Jesper Skriver 30 Sep '99

30 Sep '99

Hi, Is there a massive fiber cut, or what ? >From what I can see both above and bbnplanet has problems between the east and west cost ... traceroute to www.above.net (207.126.96.163), 30 hops max, 40 byte packets 1 eth9-0-0.sltnxt2.ip.tele.dk (193.162.159.76) 0.852 ms 0.579 ms 0.521 ms 2 feth5-0-0.sltnxt1.ip.tele.dk (195.249.2.26) 0.775 ms 0.789 ms 0.664 ms 3 atm1-0-3.kd4nxi1.ip.tele.dk (195.249.2.21) 4.470 ms 3.098 ms 4.514 ms 4 ser4-0.45M.linxnxi1.ip.tele.dk (195.41.41.246) 23.299 ms 23.213 ms 25.149 ms 5 linx.london.above.net (195.66.224.76) 38.534 ms 71.093 ms 26.651 ms 6 nyc-lhr-stm-1.nyc.above.net (209.249.0.106) 92.307 ms 91.116 ms 91.575 ms 7 iad-nyc-oc12-1.iad.above.net (209.249.203.49) 94.735 ms 94.857 ms 96.739 ms 8 pao-iad-oc3.pao.above.net (207.126.96.145) 1072.156 ms 1054.131 ms * 9 * sjc-pao-oc12-2.sjc2.above.net (207.126.96.66) 1069.846 ms * 10 main1-core5.sjc.above.net (209.133.31.134) 669.436 ms 687.632 ms 680.923 ms 11 www.above.net (207.126.96.163) 663 ms 676.565 ms 682.928 ms traceroute to cio-sys.cisco.com (192.31.7.130), 30 hops max, 40 byte packets 1 eth9-0-0.sltnxt2.ip.tele.dk (193.162.159.76) 1.101 ms 0.544 ms 0.505 ms 2 atm1-0-72.kd4nxi2.ip.tele.dk (195.249.7.89) 5.675 ms 5.629 ms 5.967 ms 3 atm1-0-76.boanxt1.ip.tele.dk (195.249.7.106) 11.058 ms 9.940 ms 9.957 ms 4 feth4-0-0.boanxi2.ip.tele.dk (195.249.5.202) 11.139 ms 10.210 ms 10.337 ms 5 pos1-0-0.155M.NewYorknxi2.ip.tele.dk (195.249.5.190) 88.229 ms 86.665 ms 86.373 ms 6 s4-0-1.nyc4-cr5.bbnplanet.net (4.1.78.5) 94.553 ms 92.325 ms 87.544 ms 7 p3-3.nyc4-nbr3.bbnplanet.net (4.0.3.162) 87.342 ms 86.945 ms 86.760 ms 8 p2-1.chicago1-nbr1.bbnplanet.net (4.0.3.121) 1213.229 ms 1243.234 ms 1244.851 ms 9 p2-1.paloalto-nbr2.bbnplanet.net (4.24.7.18) 1560.060 ms 1594.888 ms 1578.081 ms 10 p0-0-0.paloalto-cr18.bbnplanet.net (4.0.3.86) 1557.100 ms 1586.801 ms 1576.663 ms 11 h1-0.cisco.bbnplanet.net (4.1.142.238) 1590.069 ms 1657.856 ms 1591.913 ms 12 sty.cisco.com (192.31.7.1) 1572.650 ms 1573.448 ms 1570.231 ms 13 cio-sys.cisco.com (192.31.7.130) 1570.323 ms 1642.216 ms 1608.919 ms /Jesper -- Jesper Skriver (JS4261-RIPE), Network manager Tele Danmark DataNet, IP section (AS3292) One Unix to rule them all, One Resolver to find them, One IP to bring them all and in the zone to bind them.

11 10

backhoe kiddies
by Dan Hollis 30 Sep '99

30 Sep '99

It's awesome the damage an individual with a backhoe can do to the net. With one flip of a lever, kill an entire continent. http://www.internettrafficreport.com/cgi-bin/tr_chartpage.pl?NorthAmerica -Dan

1 0

Re: Massive fiber cut??
by Sean Donelan 30 Sep '99

30 Sep '99

<spork(a)inch.com> writes: > (Sean, where are you??) Not in Ohio. One of the things I couldn't take with me when I changed jobs was my network contacts database. I need to start collecting phone numbers and contacts all over again.

1 0

issues with DoC, ICANN, and NSI resolved
by Ravi Pina 28 Sep '99

28 Sep '99

"The resolution of issues among the Department of Commerce, the Internet Corporation for Assigned Names and Numbers (ICANN), and Netowrk Solutions, Inc. (NSI) is embodied in five agreements (and the appendices to those agreements): . . ." http://www.ntia.doc.gov/ntiahome/domainname/agreements/summary-factsheet.htm -r -- echo "send pgp key" | mail ravi(a)cow.org "Begin to eat the dust.. the... to eat the... to eat the... the... to eat the dust... the dust of... Baja California."

2 1

Re: FW: your mail
by rs＠seastrom.com 27 Sep '99

27 Sep '99

> > > I have listened to their seminar about this... As the simple L5 firewall > > > it's not bad, through it realise the fixed set of ruls and defends your > > > from the simple SMTP attacks only. But anyway, IOS FW is just what 90% of > > > the customers need... > > > > How would IOS FW perform on Cisco 7x00-class equipment with 100M-to-Gigabit > > traffic ? > > Umm... Very poorly. At the low end it's acceptable. Gigabit traffic sucks on 7500 series routers even without any kind of filtering. The 7000-series routers, if they have an SSE, will do standard and extended access lists in the switch engine. Now, given the limitations of CX-FEIP-2TX boards (the only faste boards that will work in a non-RSP 7000), you are lucky to get 70 mbit/sec through that. If you have fddi, you can get most of the way to 100 mbit/sec one way (the CX-FIP cards, which are the only FDDIs that work in a 7000, won't do full-duplex). The 7500-series routers, you really want to get a VIP2-50 rather than a 2-40 or lower if you're going to be doing filtering on the linecard. You can load the fast ethernets up just fine there. 400 mbit/sec seems to be the upper limit of the currently shipping generation of gigE cards for the 7500 series. Hope this helps (and standing by for corrections from the #cisco IRC mafia...) ---Rob

5 5

Re: IP bandwidth usage collector?
by Robert Gash 27 Sep '99

27 Sep '99

Jim Thompson wrote: > > Shall I be the 300th person to tell you about mrtg? You're pretty darn close actually. For the list, I'll be writing a little script that will pull these values from SNMP and store them in a SQL database, so the admin script as well as the user can check their bandwidth charts (I do use MRTG, but I like being able to look at one screen and see the exact amount of traffic that my customers have transferred to the net as well as what they owe us (if anything). I'll hopefully start work on it tomorrow afternoon and might have something finished by the end of the week. I'll let the list know where they can grab their copy. -- __ ___ __ Robert Gash /\ \ /\_ \ /\ \__ __ __ ____\ \ \___ __ \//\ \ ___\ \ ,_\ /'_ `\ /'__`\ /',__\\ \ _ `\ /'__`\ \ \ \ / __`\ \ \/ /\ \L\ \/\ \L\.\_/\__, `\\ \ \ \ \/\ \L\.\_ \_\ \_/\ \L\ \ \ \_ \ \____ \ \__/.\_\/\____/ \ \_\ \_\ \__/.\_\/\____\ \____/\ \__\ \/___L\ \/__/\/_/\/___/ \/_/\/_/\/__/\/_/\/____/\/___/ \/__/ /\____/ email: gashalot(a)gashalot.com 13356305 :icq \_/__/ url: http://www.gashalot.com 603.853.8389 :fax

3 2

bgp design
by Robin DeFreitas 27 Sep '99

27 Sep '99

Does anyone know of any good books that specifically deal with BGP design !? I have a large OSPF network and plan to implement 4 internet access points at different parts of the network. I want to control traffic from certain areas to specific internet access points, but have redundancy to other internet access points in case the primary route is down. I was looking at implementing either route maps with attributes to prefer a certain path over the other or break my AS into Confederations. Is this a solid design plan or is there a better way of doing this ?

4 5

Re: FW: your mail
by owen＠dixon.DeLong.SJ.CA.US 27 Sep '99

27 Sep '99

Since they are VIP PA's, the problem is PCI bus limitations. Owen > > On 25 Sep 1999, Robert E. Seastrom wrote: > > > > 400 mbit/sec seems to be the upper limit of the currently shipping > > generation of gigE cards for the 7500 series. > > <scratching head> > > 400 mbit != 1 gigabit > > I guess the marketing department wins again.... > > /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ > Patrick Greenwell > "This is our time. It will not come again." > \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/ > > >

4 3

SUMMARY: DACS vendor
by Peter Polasek 27 Sep '99

27 Sep '99

================================= SUMMARY OF DACS VENDOR RESPONSES: ================================= Original Message: ----------------- We are considering using Digital Access Cross Connect units (DACS) to split T1 local loops into fractional T channels. Given that our current knowledge of DACS does not extend significantly beyond the ability to spell the acronym, I am soliciting advice from the network group for recommended hardware vendors. Cost is not an object, we are primarily concerned with choosing the 'industry standard' manufacturer with the highest reliability and knowledgeable technical support. Please share your experiences (positive or negative) with DACS equipment. In an effort to minimize traffic on the NANOG list, I request that responses be sent to my e-mail address and I will submit a summary when all the votes are in. The vendors under consideration now are Adtran, Eastern Research, StarDAX, and Fisher Price (we have some reservations about the latter product). We will be splitting T1's into 128's, 256's, ... (not 56Kb lines because the lower bandwidth lines have 56Kb local loops). The T1 lines would typically be split into a maximum of eight various speed ports (128-512, occasionally 56, but not typically). I am assuming that products exist to split directly into a 256Kb without requiring that multiple 56's to be bundled, but this may not be the case (I would appreciate clarification on this point). We may subsequently use it for 56K lines, so a chassis that supports both 8-way and 24-way cards is helpful. Approximately 50 T1's are currently candidates for splitting to lower bandwidth. Our preference is to use multiple small capacity DACS rather than larger 'megaDACs' to minimize the impact of a chassis failure. We will have at least one unpopulated hot spare in each location. Response Summary: ----------------- Much to my surprise, there does not appear to be a single dominant DACS vendor. It seems that DACS manufacturers can do no wrong because nearly all respondents were quite satisfied with the product they had selected. The most popular device appears to be the Alcatel (formerly DCS) product, followed closely by the Tellabs Titan series DACS. There was a series of other choices, listed in approximate order of popularity, including NewBridge, Adtran, Lucent, Fujitsu, and Nortel. The only universally maligned product was the Fisher Price unit mentioned in my original message. Buddy Venne provided some keen insight into the issues and concerns with this vendor: btw, I would give up on FP also, the rumor is that they are about to be bought by Mattel and that will dilute the value of their networking hardware, especially the pool equipment. Aled Morris suggested using Cisco's Lightstream 1010 to channelize lower bandwidth lines through a single T1 line. This is an excellent solution when Cisco routers are deployed and all interfaces are contained within a single router. We have begun experimenting with this option in our newer Cisco interfaces; however, we still need a DACS solution for the older interfaces that span multiple routers. Peter Helmenstine provided the most succinct summary, repeated verbatim below: FCC requires the Bells to use multiple vendors like: Fujitsu, Lucent, Nortel, Alcatel, TelLabs and a few others. In my experience: USWest uses all above Level3 uses Alcatel(DS1s) NextLink uses Nortel(DS3s) and Tellabs(DS1s) Bell Atlantic primarily uses Fujitsu Bell South primarily uses Lucent PacBell primarily uses Nortel TCG/AT&T uses Lucent MFS/MCIWorldCom uses Fujisu This following book has a section describing features of different vendors: Sonet : A Guide to Synchronous Optical Network by Walter J. Goralski http://www.amazon.com/exec/obidos/ASIN/0070245630/o/qid=938033335/sr=8-2/00… One respondent cleverly constructed a cross connect unit out of ATM switches using circuit emulation cards. I don't feel quite up to this task even though I believe I once saw MacGyver perform a similar feat (though, if memory serves, MacGyver was blindfolded and had to construct the emulation cards out of chewing gum and a discarded paper clip). Final Selection: ---------------- We will evaluate three products - Alcatel, Tellabs, and Adtran. The first two are selected because they are the most popular, if not dominant, products. The Adtran is attractive for our specific environment because we have used their products extensively for many years and have been satisfied with both the products and the vendor. I would like to thank the following for their extremely helpful input (hopefully I haven't left anyone out): Erik Montemer <emontem1(a)bigred.unl.edu> Tim Hodges, Triangle Tele Coop Assoc. <ttceng(a)ttc-cmc.net>: Joe Abley <jabley(a)patho.gen.nz>: Kevin Cappis <kevin.cappis(a)shaw.ca>: Don Pace <pace(a)sprintans.net> Venne, Buddy <Buddy.Venne(a)ceridian.com> Chad Skidmore <cskidmore(a)avistacom.net> Richard Newcomb <rnewcomb(a)pathone.net> Peter Helmenstine <pete(a)internap.com> Paul E. Erkkila <pee(a)globalcenter.net> Wilson, Chris <CHRIS.WILSON(a)BELLSOUTH.COM> Rob Russo <rrusso(a)bbn.com> Aled Morris <aledm(a)routers.co.uk> Dave Cooper <dcooper(a)gulp.org> Andy Ellifson <andy(a)mx1.thelocation.com> Administrator <admin(a)nv.net> Ron Buchalski <rbuchals(a)hotmail.com> Peter Polasek <pete(a)cobra.brass.com> Gant, Patrick <patrickg(a)info-prod.com> Alex Rubenstein <alex(a)nac.net> Jim Hart <jimh(a)planbcom.com> I have included relevant excerpts from all responses below for the enthusiastic reader. Thanks, Peter Polasek pete(a)brass.com <--- begin included responses ---> From: "Erik Montemer" <emontem1(a)bigred.unl.edu> ----------------------------------------------- I'm not a telco DACS backbone guru. My experiences with Tellabs, has been pretty positive, the ease of use when doing frame cuts, made for less downtime, and less work. The Tellabs 1/0 DACS also was easy to configure for circuit simulation when testing different types of scenarios before deploying various services. I've only had experience with Tellabs 1/0 (ds-1 to ds-0) DACS, and Tellabs 3/1 DACS (DS-3 to DS-1). As far as industry standardization is concerned, most vendors should be compliant with ITU, ANSI T1 standards, as well as their accompanying signalling standards. I honestly don't think there's one specific DACS used as an exemplary standard. It all depends on who's purchasing the DACS, how convincing the sales rep is, and/or if upper management considers bottom line over delivering quality service. I would say a majority of telcos use Tellabs, or Adtran, but I could be wrong. Hope this helps. From: "Tim Hodges, Triangle Tele Coop Assoc." <ttceng(a)ttc-cmc.net>: ------------------------------------------------------------------- I work at a independent rural telephone company. We replaced our CO DACS last year with one made by DCS. They have since been bought out by Alcatel. The general terminology for DACS are like 3/1/0 for a box that will take DS3/T-1's/DS0 for example. The one we bought is a 3/1. We have been pretty happy with it, but it is a pricey product. From: Joe Abley <jabley(a)patho.gen.nz>: -------------------------------------- We build a digital cross connect out of a Stratacom BPX ATM switch using circuit emulation cards. We found that we could build a unit with equal capacity to that available on our previous DCC for much, much less money (plus we get a useful box to play with once all the legacy circuit-based services are finally banished). Just something you might like to think about (it worked for us). From: Kevin Cappis <kevin.cappis(a)shaw.ca>: ------------------------------------------ Check out Newbridge if cost is not a factor. They have been around for quite awhile, and have a good reputation. For something a little cheaper (less features), but good and reliable, look into Alcatel. They have a nice 1/0 mux that's not that expensive. From: Don Pace <pace(a)sprintans.net> ----------------------------------- Take a serious look at the tellabs DACS equipment . It is being used by many a phone company. They sell a real nice 8 port T1 unit that uses the same cards as some of their larger units. They have units that that grow as big as you want them. The craft guys around here love them. From: "Venne, Buddy" <Buddy.Venne(a)ceridian.com> ----------------------------------------------- I would greatly appreciate seeing the final summary, I may need to DACS some T1's for fault-tolerance reasons, if one T1 were to go down, a back up line can be moved in place. As well, I could use the lines as test lines during non-production hours. btw I would give up on FP also, the rumor is that they are about to be bought by Mattel and that will dilute the value of their networking hardware, especially the pool equipment. From: Chad Skidmore <cskidmore(a)avistacom.net> --------------------------------------------- One of the recognized leaders in DACS equipment is Tellabs. Take a look at their Titan 532 at http://www.tellabs.com if you are just looking for 1/0 DACSing. From: Richard Newcomb <rnewcomb(a)pathone.net> -------------------------------------------- However both Newbridge and Alcatel offer this product, with Alcatel having a wide variety of options. From: Peter Helmenstine <pete(a)internap.com> ------------------------------------------- This book has a section describing features of different vendors: Sonet : A Guide to Synchronous Optical Network by Walter J. Goralski http://www.amazon.com/exec/obidos/ASIN/0070245630/o/qid=938033335/sr=8-2/00… FCC require the Bells to use multiple vendors like: Fujitsu, Lucent, Nortel, Alcatel, TelLabs and a few others... In my experience: USWest uses all above Level3 uses Alcatel(DS1s) NextLink uses Nortel(DS3s) and Tellabs(DS1s) Bell Atlantic primarily uses Fujitsu Bell South primarily uses Lucent PacBell primarily uses Nortel TCG/AT&T uses Lucent MFS/MCIWorldCom uses Fujisu From: "Paul E. Erkkila" <pee(a)globalcenter.net> ---------------------------------------------- Titan. 0-1 for ds1->frac 3-1 for DS3->ds1. We've got hundreds of them in place. From: "Wilson, Chris" <CHRIS.WILSON(a)BELLSOUTH.COM> -------------------------------------------------- The telco industry generally standardizes on Alcatel, Lucent, and Tellabs. I've heard more and more negative feedback about the Tellabs products laterly. I'd say of the three, Alcatel is the way to go. From: "Rob Russo" <rrusso(a)bbn.com> ---------------------------------- Adtran is one of the better ones... have you checked out CAC (Carrier Access Corp)? I am not sure if they have this, but they have other device with built DACS.... From: Aled Morris <aledm(a)routers.co.uk> --------------------------------------- The Cisco Lightstream 1010 can do this using the T1 channel cards. (I assume you want a T1 trunk split into two or more T1 tribs with N timeslots cross connected, if you want V.35 presentation of the tribs then ignore this message!) Although it isn't strictly a DACCS, you might feel more comfortable supporting a Cisco device depending on what else you have in your network. From: Dave Cooper <dcooper(a)gulp.org> ------------------------------------ Depending on what kind of capacity you are talking about, the industry standard for 1/0 DACS' is the Alcatel. however, for the smaller/less volume, i have heard that the newer StarDAX's are pretty nice. From: Andy Ellifson <andy(a)mx1.thelocation.com> ---------------------------------------------- AT&T Local Services uses the Lucent product for DS0 DACS. From: Administrator <admin(a)nv.net> ---------------------------------- I have had excellent luck with the Adtran Atlas 800 product in several sites over the past year. It has just the right mix of port density and flexibility for our current needs and has some very slick PRI switching capabilities as well. Uptime has been 100% for ~8 months now (single PSU, 120 VAC) other than when we were upgrading software images. Speaking of that, whenever you load a new image (to the main unit or a module) you can schedule the reboot for a later time--allowing you to do the upgrade while you're in the office and have it restart at 0200 for example. The user interface is intuitive and well-crafted--wish I could say the same about some of our other vendors' interfaces. The tech support comes from a special group dedicated to/trained on the Atlas line and they have always answered my calls or returned them within an hour. They even got us a custom SW load to deal with a new ISDN parameter in less than a week. I'd gladly buy (and plan to) more of them. From: "Ron Buchalski" <rbuchals(a)hotmail.com> -------------------------------------------- A simpler solution which would require less space, power, money, wire pairs, and administration for your group: 1) Provision a few Channelized DS3s into your LEC network 2) Install Channelized DS3 into your router 3) Let the LEC be responsible for owning/managing the DACS, and provisioning the FT1s/T1s onto the T1s within the channelized DS3s. You'll save the space and power normally required for banks of T1 CSUs and the DACS, and you won't need to manage the crossconnects for the DACS. Circuit-wise, it should cost you less to provision your T1s into your facility via a few DS3s rather than fifty T1s, and you won't eat up your incoming telco wire facilities quickly. From: Peter Polasek <pete(a)cobra.brass.com> From: "Gant, Patrick" <patrickg(a)info-prod.com> ---------------------------------------------- Look into the Adtran ATLAS 800 product. From: Alex Rubenstein <alex(a)nac.net> ------------------------------------ TadIran is making some really nice, yet inexpensive stuff. Also, TelInc is making one that is really cheap, but only handles four DS3's; perfect for the remote pop. I'm thinking of getting one of those. Something like $12k for a 3xds3, 30xds1 system. From: "Jim Hart" <jimh(a)planbcom.com> ------------------------------------ You might want to consider the DACS-II from Lucent the AX/4 or AX/11 (4 and 11 Slots models). We have a need for doing something similar and this seems that it would more than fit the bill. http://www.lucent.com/dns/products/dacs.html

4 3

IP bandwidth usage collector?
by gashalot＠gashalot.com 26 Sep '99

26 Sep '99

This might be a little off topic, but I run a fairly large network, and we sell dedicated servers to clients. All of the dedicated machines are connected to our network of Cisco 2924XL-M Catalyst switches. I would like to be able to collect bandwidth usage statistics for their ports that way I can easily bill them for overage, and I don't want to have to install anything on the machines themselves to remove the threat of tampering. Does anyone know of a package that does this? Or would it just be best to write some quick Perl hack to pull the bytecounter out of the Cisco's SNMP information? Thanks for any help anyone might be able to provide. -- Robert Gash System Administrator, FastURL Personal Email: gashalot [at] gashalot [dot] com

3 2