March 1996 - Test - lists.nanog.org

Re: MCI [ATM overhead]
by Paul Ferguson 19 Mar '96

19 Mar '96

At 07:51 PM 3/19/96 -0800, Jim Forster wrote: > >There's beginning to be some expectation that there will be a transmission >capacity crunch in the carrier's Sonet nets, and this ~25% ATM cell tax may >be looked at carefully as packet over Sonet solutions emerge. > I think it will be critically important, especially in long-haul interconnect scenarios. Historically speaking, as providers have replaced congested links with faster ones, they have also become quickly congested. And since these high-speed long-haul links are not cheap, you can bet that these organizations want to get the most bang-for-the-buck. I can't imagine that ~25% bandwidth overhead would be acceptable, especially when more efficient alternatives exist. Nice synopsis, Jim. - paul

1 0

Re: MCI [ATM overhead]
by avg＠postman.ncube.com 19 Mar '96

19 Mar '96

Jim Forster <forster(a)cisco.com> wrote: >He's talking about the overhead due to carrying variable length IP packets >in fixed length ATM cells. ... >There's beginning to be some expectation that there will be a transmission >capacity crunch in the carrier's Sonet nets, and this ~25% ATM cell tax may >be looked at carefully as packet over Sonet solutions emerge. Given the bimodailty of IP traffic size distribution (about 40% of packets are small, like TCP ACKs or telnet/rlogin keystrokes) the ATM "cell tax" is closer to 32%. I.e. a dual clearline DS-3 actually carries as much user data as OC-3c ATM. Which, incidentally, was why SprintLink backbone design is easily expandable to dual links (that includes carefully considering implications for routing). Sean presented that design on NANOG a year ago, BTW. Funny thing, the design is expandable beyond that, too, so OC-3 ATM is already obsolete. --vadim

1 0

Re: MCI [ATM overhead]
by jogden＠merit.edu 19 Mar '96

19 Mar '96

If we are going to talk about ATM overhead when doing TCP/IP don't we need to talk about the overhead of partially filled ATM cells? Won't that cost you about 1/3 of your available bandwidth? -Jeff Ogden Merit

4 3

Re: MCI
by rcraig＠cisco.com 19 Mar '96

19 Mar '96

At 12:29 96/3/19, Wolfgang Henke wrote: ... >I have assembled a list of several best effort packet routers including >known research efforts: > >http://www.whnet.com/wolfgang/giga.html The list shows the 75xx as doing about 80mbps, which is incorrect. I've been doing extensive testing in a realistic testbed with a three HSSI, two FDDI configuration, and I get 160mbps without difficulty. The aggregate throughput at the moment is limited, not by the unit under test, but by the generators I'm using. I've just acquired some more powerful ones, so should be able to stress the box a bit more. The results I anticipate will approach Product Marketing's numbers ;-) Robert.

1 0

Re: MCI
by Paul Ferguson 19 Mar '96

19 Mar '96

At 12:54 PM 3/19/96 -0500, Gordon Cook wrote: > >Arun Welch wrote: > >Yup. There's work in progress to put IP directly on SONET, which should save >people with OC-3c backbones from paying the ATM penalty. > >Cook: would you please give a pointer to whom they are and what one reads >or looks at to get up to speed? (ie IP over sonet) > RFC-1619, 'PPP over SONET/SDH'. - paul >********************************************************************* >Gordon Cook, Editor & Publisher Subscriptions: Individ-ascii $85 >The COOK Report on Internet Individ. hard copy $150 >431 Greenway Ave, Ewing, NJ 08618 USA Small Corp & Gov't $200 >(609) 882-2572 Corporate $350 >Internet: cook(a)cookreport.com Corporate Site Lic. $650 >http://pobox.com/cook/ for new COOK Report Glossary of Internet terms >********************************************************************* > -- Paul Ferguson || || Consulting Engineering || || Reston, Virginia USA |||| |||| tel: +1.703.716.9538 ..:||||||:..:||||||:.. e-mail: pferguso(a)cisco.com c i s c o S y s t e m s

1 0

Re: MCI
by bajaj＠bellcore.com 19 Mar '96

19 Mar '96

>> > > > > What kinds of routers can route at near wire speeds a bunch of ports at > > > > 155Mb ATM? > > > > > > None right now. Gigarouter can handle it theoretically, but... > > > > > > I'm being picky, but it's a moot question since doing wirespeed OC3c using > > > ATM is impossible in of itself. Hell, mapping of > > > SONET framings to ATM cells take off about 6 mbps right there. > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > By the way, I did not post the above.. Oops. Replied to the right thread but wrong author. > bajaj(a)bellcore.com (Shikhar Bajaj) wrote: > > A couple of minor corrections here. ATM cells are mapped into SONET frames and > > not the reverse. Also, I presume that the 6 Mbps that you are talking about is that > > the ATM bit rate over an OC-3 is 149 Mbps as opposed to the OC-3 pipe > > rate of 155 Mbps. Those 6 Mbps are taken up by the SONET management overhead > > (section, line, and path) in the frame. This is independent of whatever goes in the > > SONET payload envelope and has nothing to do with ATM. > > > It seems to me that Sonet OC-3 is often taken as transport for 3 DS3s, > which would reduce 155.530 Mbps to 134.208 Mbps (3 x 44.736 Mbps), less > than the 149 Mbps you cite above. The 149 Mbps assumes that you have a native ATM/OC-3 interface where the SONET transport is STS-3c ('c' for concatenated). Here, a single connection can use the full bandwidth of the pipe. In your example, a DS-3 gets mapped into an STS-1 and then 3 STS-1s get muliplexed into an STS-3. The multiplexing is for trunking purposes and there is no relation between any of the STS-1's (other than they are byte multiplexed). Thus, the maximum bandwidth available for any single connection is 51 Mbps (or DS-3 in this case). Routers with ATM/OC-3 interfaces do STS-3c framing. > Do you have a good recommendation for Gordon Cook, TCP/IP direct over Sonet? I've heard rumblings about such work going on but I am not personally familiar with any of it. The only thing I can think of is some work that the IETF did for PPP over SONET (RFC 1619). I *think* it became an Internet Standard. Shikhar

1 0

Re: Allocation of IP Addresses
by Curtis Villamizar 19 Mar '96

19 Mar '96

In message <199603150738.CAA11310(a)jekyll.piermont.com>, "Perry E. Metzger" writ es: > > Curtis Villamizar writes: > > Hoarding is only effective if there is a scarce resource, even if it > > is artificially scarce, as in the two "gas shortages" of the 1970s > > (when there were huge gas company stockpiles and enormous gas company > > profits). > > The gas shortages of the 1970s were caused by the fact that the > U.S. government regulated oil prices. Whenever you artificially lower > the price of anything, you create shortages. When decontrol occurred, > shortages mysteriously and totally stopped. Amazing, ain't it. Perry, Just a point of history. The gas shortages were both triggered by OPEC moves which reduced crude oil imports. There were enormous oil reserves in all of the major oil companies. They were not willing to release those oil reserves because of their LIFO inventory practices, used to defer tax liability. The oil companies tried to convince the government to waive the taxes that were due if they sold down their inventory that was on the books at oil prices going back many decades. The govenment would not allow them to convert from LIFO to FIFO inventory and simply waive taxes on the billions of dollars of oil on the books for almost no value so the oil companies tried holding the consumer hostage in hopes of putting enough pressure on. There were no regulations on gasoline prices and they did go up but oil companies were hesitant to raise them too high and seem to be price gouging during a fabricated shortage but also didn't want to pay taxes on the profit of selling at the current price with inventory at the old prices. Curtis

1 0

Re: MCI
by bajaj＠bellcore.com 19 Mar '96

19 Mar '96

> > > What kinds of routers can route at near wire speeds a bunch of ports at > > 155Mb ATM? > > None right now. Gigarouter can handle it theoretically, but... > > I'm being picky, but it's a moot question since doing wirespeed OC3c using > ATM is impossible in of itself. Hell, mapping of > SONET framings to ATM cells take off about 6 mbps right there. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ A couple of minor corrections here. ATM cells are mapped into SONET frames and not the reverse. Also, I presume that the 6 Mbps that you are talking about is that the ATM bit rate over an OC-3 is 149 Mbps as opposed to the OC-3 pipe rate of 155 Mbps. Those 6 Mbps are taken up by the SONET management overhead (section, line, and path) in the frame. This is independent of whatever goes in the SONET payload envelope and has nothing to do with ATM. Shikhar

2 1

Re: MCI
by David Miller 19 Mar '96

19 Mar '96

On Mon, 18 Mar 1996 Jasper(a)cloud9.net wrote: > MCI issued a press release this afternoon. Two items of interest: > > 1) MCI is upgrading its backbone from 45 megabits to 155 megabits per > second. They expect this to be done by mid April. I think this has been > coming along for a while now. > A question: > > Does anyone know when Sprint and Alternet will have their backbones up to > 155 megabits per second. A guy at MCI told me today they wouldn't have it > done until 1st quarter of 1997. Another question: What kinds of routers can route at near wire speeds a bunch of ports at 155Mb ATM? --- David Miller ---------------------------------------------------------------------------- It's *amazing* what one can accomplish when one doesn't know what one can't do!

4 4

Re: MCI
by Wolfgang Henke 19 Mar '96

19 Mar '96

> What kinds of routers can route at near wire speeds a bunch of ports at > 155Mb ATM? None right now. Gigarouter can handle it theoretically, but... I'm being picky, but it's a moot question since doing wirespeed OC3c using ATM is impossible in of itself. Hell, mapping of SONET framings to ATM cells take off about 6 mbps right there. Anyway, until then you have a router that can do this, there always are ATM switches + cluster of routers. True, Netstars's Gigarouter is the fastest one shipping now. Its aggregate advertised bandwidth is 16 Gbps using a 16x16 matrix of non-blocking crosspoint switches. I have assembled a list of several best effort packet routers including known research efforts: http://www.whnet.com/wolfgang/giga.html Corrections and additions are welcome. Wolfgang Wolfgang Henke wolfgang(a)whnet.com . http://www.whnet.com/wolfgang/ WH Networks ... ftp.whnet.com /pub/wolfgang 2672 Bayshore Parkway Suite 503 ....... +1 415 390 9316 Mountain View CA 94043 ................... fax +1 415 390 9317

1 0