May 1996 - Test - lists.nanog.org

RFC 1597
by David J. Sharp 23 May '96

23 May '96

I need a few class Cs. Hope nobody will mind if I take 192.168.0.0/20. ;) *> 192.168.22.0 144.228.71.5 0 1239 1800 1804 1128 1955 3337 ? *> 192.168.100.0/22 144.228.71.5 0 1239 1794 ? *> 192.168.216.0 144.228.71.5 0 1239 1800 1755 1273 ? Shame on you 3337, 1794 and 1273. - david

3 2

Re: RFC 1597
by Sean Doran 22 May '96

22 May '96

| I need a few class Cs. Hope nobody will mind if I take | 192.168.0.0/20. ;) Knock yourself out. :-) | *> 192.168.100.0/22 144.228.71.5 0 1239 1794 ? This appears to be a truly ancient (before 17 Jan 1996) typo or thinko or somesuch, and has been corrected. Sean.

1 0

NetSCARF Project/USEC Code Release July 1
by William B. Norton 21 May '96

21 May '96

Hi all - In April, Merit received funding from the Resource Allocation Committee (RAC) to develop and evolve a prototype ISP Statistics Collection And Reporting Facility (NetSCARF) package. The basic idea is to make it dead easy for ISPs to collect and report data about their part of the Internet. The end result will hopefully be widely-available Internet performance data much as Merit produced during the operation of the NSFNET. The NetSCARF code is easy to install, runs automatically, and consists of four separate programs. Every fifteen minutes the collections program queries all network nodes in parallel. (This is especially important for large networks where the serial skew can affect the ability to correlate data.) Nightly these raw statistics get pre-processed (cooked). The cooked data is delivered to CGI scripts using what we think is the first Public Domain implementation of the OpStats (rfc1856) client/server model. Finally, ISP performance reports (based on the cooked data) are displayed on the web via the CGI scripts. We plan on making the source and pre-compiled executables available for each of these. SNMP Version 1 is used to query the network nodes, although the code includes support for the User Security (USEC) model for SNMPv2. ( The Routing Arbiter project will use the DES Encrytion facility, but this capability will be disabled for the public release due to export restrictions. ) During the operation of the NSFNET we found there were really only three graphs that were widely seen as useful. The first cut of the code includes only these three most popular graphs: System UpTime, Interface Uptime, and the McD's chart (total packets served) by the network. Alpha Testers? -------------- The first release of the code will be made around July 1, and we currently have a version of the code that runs on SunOS 4.1.4, BSDi, and AIX systems. We are looking for a couple of ISPs to try the code out and provide early feedback. If you operate one of these Unix varieties and have the time to work with us, please e-mail your name, daytime phone #, operating environment, and a description of your network (size, types of routers, etc.) to netscarf-team(a)merit.edu with the subject line "ALPHA TESTER". The code doesn't require much - just disk space, a network attachment, and a few CPU cycles periodically. Please see me at the NANOG if you have any questions or suggestions. More information about the NetSCARF project (including the text of the proposal) is available on-line: http://home.merit.edu/~wbn/RAC/play.html Bill ---------------------------------------------------------------------------- ----- William Norton <wbn(a)merit.edu> (313) 936-2656

1 0

Re: Unregistered addresses [Re: Map Frame Relay Question]
by Erik-Jan Bos 21 May '96

21 May '96

Per, > This is the second time in just a couple of weeks I've seen mail from > people apparently believing that unregistered addresses are supposed > to work in the Net. Has the Net come so far that ignorance and > cluelessness is so widespread among people trying to manage boxes > that they think officially allocated addresses is a refinement? > A kind of dotting of i's and crossing of t's, not really necessary? > Like cars drive just fine without licence plates? > > Heaven knows there's enough people who think the whole thing works by > some strange magic, but the current apparent trend is verging on the > absurd. Don't you (only sometimes) think that the number of weekend drivers on the Internet-road is increasing, just like the number of prefixen in the default free routers of the Internet (hint for weekend drivers: these are the routers that have a clue)? BTW, see ftp://ftp.nic.surfnet.nl/surfnet/net-management/ip/nets.ps for the latest update on the number of prefixen. :-). __ Erik-Jan.

1 0

Re: PPP over SONET
by Peter Lothberg 21 May '96

21 May '96

> On Tue, 21 May 1996, Peter Lothberg wrote: > > > Or do you have many small OC-3 routers? > ^^^^^^^^^^^^^^^^^^ > Now, that's a thought! ;-) > > Michael Dillon Voice: +1-604-546-8022 > Memra Software Inc. Fax: +1-604-546-3049 > http://www.memra.com E-mail: michael(a)memra.com > > Exponential growth, yes, well where do you think circiuts like this will be before the end of the year? Hssi2/0 is up, line protocol is up Hardware is cxBus HSSI Description: PL00190596 ICM-PEN-2<-->ICM-UK-1 Internet address is 198.67.131.25/29 MTU 4470 bytes, BW 45045 Kbit, DLY 200 usec, rely 255/255, load 248/255 Encapsulation HDLC, loopback not set, keepalive set (10 sec) Last input 00:00:00, output 00:00:00, output hang never Last clearing of "show interface" counters 5d02h Output queue 0/40, 24883016 drops; input queue 0/75, 175 drops, 109723 flushes 5 minute input rate 21531000 bits/sec, 15203 packets/sec 5 minute output rate 43820000 bits/sec, 13082 packets/sec -Peter

1 0

Unregistered addresses [Re: Map Frame Relay Question]
by Per Gregers Bilse 21 May '96

21 May '96

This is the second time in just a couple of weeks I've seen mail from people apparently believing that unregistered addresses are supposed to work in the Net. Has the Net come so far that ignorance and cluelessness is so widespread among people trying to manage boxes that they think officially allocated addresses is a refinement? A kind of dotting of i's and crossing of t's, not really necessary? Like cars drive just fine without licence plates? Heaven knows there's enough people who think the whole thing works by some strange magic, but the current apparent trend is verging on the absurd. >I am working at a site that has a Cisco 7010? Router >running V 10.0 The router uses a frame relay connection >to an Internet Provider to connect to the outside World. > >Most of the subnets here are "legal" (i.e. registered w/Internic). >However, we have several that aren't. > >Everything works great of the legal subnets. However, the >'illegal' ones cannot connect or even ping anything on the >Frame Relay side of the router. There are technical reasons >why we are not able to get the bogus subnets changed right now. > >I'm wondering if I need to have a map frame relay directive >in the interface description to handle traffic to & from >the subnet that's giving me problems. > >The access-list filters look OK and the interface in question >has no filtering at all. > >Ideas would be greatly appreciated! -- ------ ___ --- Per G. Bilse, Mgr Network Operations Ctr ----- / / / __ ___ _/_ ---- EUnet Communications Services B.V. ---- /--- / / / / /__/ / ----- Singel 540, 1017 AZ Amsterdam, NL --- /___ /__/ / / /__ / ------ tel: +31 20 6233803, fax: +31 20 6224657 --- ------- 24hr emergency number: +31 20 421 0865 --- Connecting Europe since 1982 --- http://www.EU.net e-mail: bilse(a)EU.net

1 0

Re: PPP over SONET
by Peter Lothberg 21 May '96

21 May '96

> But you can configure PVC's with ATM, working with an ATM > cloud, similar to frame relay, no? And if this is the case, then > a large backbone-type provider may desire to set up an ATM cloud, > making it easy to add POP's (just add them to the cloud), and allow > for more layer 2 routing (just configure PVC's between every POP you > have, building a full web, and a traceroute would now travel 2 or 4 > hops instead of 12...?) If the number of hops that shows up in a traceroute is a design parameter when building a network. And that is concidered higher than the overall performance, yes.. IP over IP tunnels would also solve the problem -:) (or disable loop detection by taking out the TTL decrement -:) > instead of some sort of hierarchichal star topology, you'd > have a large web... Your ATM network topology will still likely be looking like what your IP network would have looked, right? In order to make efficient use of your cloud, all routers has to be neighbours to all other routers, so in practice there is a limit to the number of routers you can have. Many large IP-networks are still built as logical star networks. Depending on the size of your IP network, a lot of traffic will be in/outbound to other ISP's, and here you need a router for various reasons. > and one more potential benefit to ATM -- won't it scale > very well to OC-12? ie, when you want to upgrade your network to > OC-12, you'd have no changes to make? Asuming everyone talks to everyone, but, what router do you connect at the edge that speaks OC-12? Or do you have many small OC-3 routers? --Peter

2 1

Solaris retransmission bug eats Internet links
by Hank Nussbacher 21 May '96

21 May '96

If you can think of a better list to post this, please forward it... The following TCP bug has been measured to consume as much as 15% of Internet links. ISPs who run heavily congested links and/or satellite links will lose 15% of their bandwidth as we have due to this error by Sun. Please distribute this announcement to those running Solaris 2 as often as possible. Thanks, Hank Nussbacher IBM Israel -------------------------------------------------------------------- During the last few weeks we were conducting a survey of our internet communication performance. As a result of the survey we have discovered that your host is involved in a large portion of TCP/IP retransmissions. This phenomena significantly decreases the communication performance. If your host is a Sun running Solaris 2, than we think we know what is the source of the problem. It appears that Solaris 2 has a bug in the TCP/IP implementation, causing excessive retransmissions to occur. The following is an excerpt from the May 3rd Solaris 2 FAQ: ----------------------------------------------------------------------------- +5.49) Solaris 2.5 and Solaris 2.4 patch 101945-34+ have poor TCP performance over slow links. Solaris 2.5 and Solaris 2.4 kernel patch 101945-34 and later have a bug in their TCP retransmission algorithm that cause excessive retransmissions over slow links, Sun's bug ID is #1233827. A work around for this bug is running the following commands at system boot, e.g., by adding them to /etc/init.d/inetinit (values are in milliseconds): /usr/sbin/ndd -set /dev/tcp tcp_rexmit_interval_min 3000 /usr/sbin/ndd -set /dev/tcp tcp_rexmit_interval_initial 3000 ----------------------------------------------------------------------------- Please note that the above mentioned patch (101945-34) included a "fix" to the retransmission problem, but apparently, the fix is not good enough (see "http://www.sun.com/sunworldonline/common/cockcroft.letters.html" section: "How can I improve my Web server's http performance?"). To summarize, it seems that all Solaris 2 versions suffer from this problem! We recommend that you apply the above workaround until the problem is resolved in a future Sun release. Please note that the problem affects mainly your host and users accessing it over large latency links, such as satellite links (as we do), or over slow links (e.g. dialup lines) or a congested network . We will appreciate if you let us know how do you intend to proceed with this. Thanks for your cooperation, Oded Comay (comay(a)post.tau.ac.il) Operating Systems group Computing Center Tel Aviv University

1 0

PPP over SONET
by Camil Samaha/TEIR/Thomson 20 May '96

20 May '96

I've been doing a lot reading the last few days on the advantages of "PPP over SONET/SDH" vs "IP over ATM AAL5". The available information bandwidth of roughly 150 Mbps out of 155.5 Mbps with STS-3c/STM-1 makes the PPP over SONET solution very attractive. I understand that SONET facilities are widely deployed throughout the carriers' broadband infrastructures but when will it become practical to run packet data over these facilities? I am aware that Cisco is planning to provide an OC-3/STM-1 interface in the 3d quarter but what about other hardware vendors? When will carriers be ready to support this service? Are multiplexers available or planned for the near future? Are there carrier-to-carrier issues that need to be worked out? Any information on this subject is very much appreciated. Thank you. -Camil ===================================================================================== Camil Samaha Senior Network Engineer Thomson Electronic Information Resources csamaha(a)teir.com =====================================================================================

2 1

Re: PPP over SONET
by Peter Lothberg 20 May '96

20 May '96

> I've been doing a lot reading the last few days on the advantages of "PPP over > SONET/SDH" vs "IP over ATM AAL5". The available information bandwidth of > roughly 150 Mbps out of 155.5 Mbps with STS-3c/STM-1 makes the PPP over SONET > solution very attractive. Speaking SDH, it's VC4, so payload is 139.264Mbit (STS-3c SPE) > I understand that SONET facilities are widely > deployed throughout the carriers' broadband infrastructures but when will it > become practical to run packet data over these facilities? I am aware that > Cisco is planning to provide an OC-3/STM-1 interface in the 3d quarter but what > about other hardware vendors? > When will carriers be ready to support this service? It's possible today, it's a matter of tariff, any SDH/Sonet network is capable of doing this. > Are multiplexers available or planned for the near future? You can't do subrates, the rest is standard tranmission gear avaliable from a variety of vendors. > Are there carrier-to-carrier issues that need to be worked out? It depends if you are buying this as a circiut within a network or as two half circiuts. If you are obtainging two half-circiuts the two corresponding carriers have to work out the maintenence procedures. --Peter Ps: ATM, what's that?

1 0