On Mon, 5 May 1997, Robert E. Seastrom wrote: : :When Sprint initially enacted peering requirements, I asked several :UUnet employees whether they were planning to do the same. The answer :was without exception that UUnet pursues peering with all technically :qualified providers at all exchange points, because that results in :the best connectivity for UUnet customers. Edward Fang <edfang@visi.net> wrote:
Actually I think this was once one of their boasting points.
Boasting is fine until economic realities start to affect your bottom line. I always held to the opinion that stratification of Internet is inevitable. Neither it is a bad phenomenon per se -- it reflect shifing of Internet being add-on service to being a core business of facilities based telecom companies. The ultimate win will be a network which is a lot more useful, with solid and sustainable business model. --vadim
Neither it is a bad phenomenon per se -- it reflect shifing of Internet being add-on service to being a core business of facilities based telecom companies. The ultimate win will be a network which is a lot more useful, with solid and sustainable business model.
--vadim
Of course there are those that simply see the PSTN as just one more way to move IP packets about. Lets see; radio, wireless, sat-link, cable, lans, PSTN, avian-carrier, seismic-wave... lots of ways to move IP packets about. the PSTN is one, but not the only one. And its not clear there is enough capacity in the PSTN to carry all the bits about. Its not clear to me that ATM will continue to work at terabit rates.. SAR may be a bit tough to do at OC768-ish rates. And then there is all that PSTN infrastructure that will have to be replaced... push all that back on the rate payers? you bet. -- --bill
On Wed, 7 May 1997 bmanning@ISI.EDU wrote:
Its not clear to me that ATM will continue to work at terabit rates.. SAR may be a bit tough to do at OC768-ish rates. And then there is all that PSTN infrastructure that will have to be replaced... push all that back on the rate payers? you bet.
I though that ATM cell sizes were so small in order to better support real-time voice and video. We are already getting to the point where 1500 byte IP packets can be transmitted end to end in the same or less amount of time as the original ATM networks were planned to be. When data rates get this high, is there any good reason to shred packets, other than maintaining compatibility with obsolete ATM gear? Michael Dillon - Internet & ISP Consulting http://www.memra.com - E-mail: michael@memra.com The bottom line is track record. Not track tearing. Not track derailing. But pounding the damn dirt around the track with the rest of us worms. -- Randy Bush
At 02:48 PM 07-05-97 -0700, Michael Dillon wrote:
I though that ATM cell sizes were so small in order to better support real-time voice and video.
The original ATM compromise was designed to avoid screwing up echo cancellation on the last mile loops which is very sensitive to delay and hard to adjust since it is old fashioned analog hardware and not digital. That is one of the more difficult aspects of the new xDSL technology is dealing with echo cancellation and crosstalk in those crummy old two-wire circuits.
We are already getting to the point where 1500 byte IP packets can be transmitted end to end in the same or less amount of time as the original ATM networks were planned to be. When data rates get this high, is there any good reason to shred packets, other than maintaining compatibility with obsolete ATM gear?
In a word "No". However, there is no good reason to fix the length at 1500 bytes. We could set the maximum to 1500 bytes and allow these cells to be shorter. We could come up with a connectionless addressing scheme, perhaps one that the hardware manufacturers can administer. Let's be generous and say 48 bytes long. What? Oh, right, IEEE already did that with 802.3 MAC. Never mind. --Kent ~~~~~ ~~~~~ ~~~~~ ~~~~~ Note new area code ~~~~ ~~~~~ ~~~~~ ~~~~~ ~~~~ Kent W. England Six Sigma Networks 1655 Landquist Drive, Suite 100 Voice/Fax: 760.632.8400 Encinitas, CA 92024 kwe@6SigmaNets.com Experienced Internet Consulting ~~~~~ ~~~~~ ~~~~~ ~~~~~ ~~~~~ ~~~~ (If you can't reach me using 760 area code, use the old 619 instead.)
At 02:48 PM 07-05-97 -0700, Michael Dillon wrote:
I though that ATM cell sizes were so small in order to better support real-time voice and video.
The original ATM compromise was designed to avoid screwing up echo cancellation on the last mile loops which is very sensitive to delay and hard to adjust since it is old fashioned analog hardware and not digital. That is one of the more difficult aspects of the new xDSL technology is dealing with echo cancellation and crosstalk in those crummy old two-wire circuits.
Part of the story. US, Japanese and European telco manufacturers had different perceptions of 'optimal' atm cell size. Depending upon the mix voice/video/data different results were obtained; believe 32 and 64 byte were mentioned. The powers that be decided upon a compromise and chose a value 'in the middle' causing everyone to redesign their equipment. The compromise value was not a power of two. Must have been '89 or thereabouts. regards, Koen ======= ___ === Koen De Vleeschauwer, Network Engineer, ====== / / / ___ ____ _/_ ==== EUnet Communications Services BV ===== /--- / / / / /___/ / ===== Singel 540, 1017 AZ Amsterdam, NL ==== /___ /___/ / / /___ /_ ====== Tel. +31 20 5305333; Fax. +31 20 6224657 === ======= [ 24hr emergency number +31 20 4210865 ]
participants (5)
-
bmanning＠ISI.EDU -
Kent W. England -
Koen De Vleeschauwer -
Michael Dillon -
Vadim Antonov