trans-Atlantic latency?
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link. What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then. Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
On 29-jun-2007, at 1:20, Neal R wrote:
What is a reasonable latency to see on a link of that distance?
You'll want to ask whether this is one-way latency or round trip time. I'm seeing this from Amsterdam: 6 51.ae0.cr2.iad1.us.scnet.net (216.246.102.94) 82.686 ms 82.762 ms 82.808 ms 7 15.xe-1-2-0.cr1.ord1.us.scnet.net (216.246.36.246) 100.792 ms 100.830 ms 100.798 ms 8 v53.ar2.ord1.us.scnet.net (216.246.95.138) 101.051 ms 101.087 ms 101.042 ms 9 v223.aggr223.ord1.us.scnet.net (216.246.94.30) 101.301 ms 101.318 ms 101.423 ms I don't think you're going to do significantly better than this, I'd be amazed if anyone can do London - Chicago below 90 ms round trip.
On Thu, Jun 28, 2007 at 06:20:31PM -0500, Neal R wrote:
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
Chicago to London: ~3950 mi New York to London: ~3470 mi c == 186,282 mi/sec (in a vacuum) 0.66 * c == 122,946 mi/sec CHI-LON: 32.128 ms NYC-LON: 28.224 ms That is one way, absolute best case, and the cables never run quite the way you want them to. If he's looking for 40 ms RTT, he is not going to get it. If he just needs 40 ms one way to London, it is possibly doable, even from Chicago. I couldn't readily find lengths for the individual segments of TAT-14, so as a representative example, we'll use TAT-12/13. From RI to the UK: 3,674 mi. ( 3674 / (0.66 * c) ) * 1000 == 29.883 ms, doubled for 59.766 ms RTT. Real world numbers seem to suggest many carriers run between 70 and 80 ms RTT from NYC to London, and I just measured around 100 ms RTT from Chicago to a host in the UK. --msa
Neal R wrote:
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
Paging Scotty, warp factor 4 please!
What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then.
The speed of light hasn't changed much. Propagation delay alone, assuming a 3000 mile straight-line path (probably on the short side) and 0.7 velocity factor in the transport medium is around 45 milliseconds round trip. Chicago to the East coast is about another 1000 miles or 15 ms, so 60ms. is probably a bit on the low side. Serialization delay depends on bit rate and packet size, easy enough to calculate. Switching delay, probably minimal.
Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
-- Jay Hennigan - CCIE #7880 - Network Engineering - jay@impulse.net Impulse Internet Service - http://www.impulse.net/ Your local telephone and internet company - 805 884-6323 - WB6RDV
Sprint has probably the lowest latency in the industry; I use them for a Los Angeles - London IPSec VPN. Typical latency is around 140-150 ms rt (70-75 ms one-way) 40 ms RT is not possible in this reality, unless the speed of light is increased or one transimits through subspace (see Star Trek) -----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf Of Neal R Sent: Thursday, June 28, 2007 4:21 PM To: nanog@merit.edu Subject: trans-Atlantic latency? I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link. What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then. Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ... -- This mail was scanned by BitDefender For more informations please visit http://www.bitdefender.com -- This mail was scanned by BitDefender For more informations please visit http://www.bitdefender.com
I used to get about 60ms from router to router in TAT12/13 (I think) from London Telehouse to NY Telehouse. Security Admin (NetSec) wrote:
Sprint has probably the lowest latency in the industry; I use them for a Los Angeles - London IPSec VPN. Typical latency is around 140-150 ms rt (70-75 ms one-way)
40 ms RT is not possible in this reality, unless the speed of light is increased or one transimits through subspace (see Star Trek)
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf Of Neal R Sent: Thursday, June 28, 2007 4:21 PM To: nanog@merit.edu Subject: trans-Atlantic latency?
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then.
Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
-- This mail was scanned by BitDefender For more informations please visit http://www.bitdefender.com
Neal R wrote:
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then.
Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
I remember voiping over the pond, from Frankfurt, germany to New York. We had to twist asterisk to even accept the sip. Time was between 80 and 90 msec. The experienced time was higher. Roger, Over and Out with their interstallar hamradio experience could do it, but to a normal citizen it was unuseble. (dsl 1000 customer, close to Frankfurt) 1 krzach.peter-dambier.de (192.168.48.2) 2.918 ms 3.599 ms 3.926 ms 2 * * * 3 217.0.78.58 85.268 ms 85.301 ms 102.059 ms 4 f-ea1.F.DE.net.DTAG.DE (62.154.18.22) 102.092 ms 110.057 ms 126.310 ms 5 p2-0.core01.fra01.atlas.cogentco.com (212.20.159.38) 126.344 ms * * 6 * * * 7 p3-0.core01.ams03.atlas.cogentco.com (130.117.0.145) 132.262 ms 139.333 ms 147.174 ms 8 p12-0.core01.lon01.atlas.cogentco.com (130.117.0.198) 76.436 ms 76.444 ms 84.374 ms 9 t1-4.mpd02.lon01.atlas.cogentco.com (130.117.1.74) 99.840 ms 99.873 ms 107.508 ms 10 t3-2.mpd01.bos01.atlas.cogentco.com (130.117.0.185) 209.678 ms 217.428 ms 225.601 ms 11 t2-4.mpd01.ord01.atlas.cogentco.com (154.54.6.22) 233.514 ms * * 12 vl3491.mpd01.ord03.atlas.cogentco.com (154.54.6.210) 243.741 ms * * 13 * * * 14 ge-1-3-0x24.aa1.mich.net (198.108.23.241) 165.776 ms 174.752 ms 193.770 ms 15 www.merit.edu (198.108.1.92)(H!) 193.812 ms (H!) 201.863 ms (H!) 209.704 ms (colo in Amsterdam) 1 205.189.71.253 (205.189.71.253) 0.227 ms 0.257 ms 0.227 ms 2 ge-5-2-234.ipcolo1.Amsterdam1.Level3.net (212.72.46.165) 0.985 ms 0.811 ms 0.856 ms 3 ae-32-54.ebr2.Amsterdam1.Level3.net (4.68.120.126) 4.235 ms 6.575 ms 1.360 ms 4 ae-2.ebr2.London1.Level3.net (4.69.132.133) 19.097 ms 12.816 ms 18.220 ms 5 ae-4.ebr1.NewYork1.Level3.net (4.69.132.109) 78.197 ms 78.769 ms 87.062 ms 6 ae-71-71.csw2.NewYork1.Level3.net (4.69.134.70) 78.068 ms 79.058 ms 89.192 ms 7 ae-22-79.car2.NewYork1.Level3.net (4.68.16.68) 142.665 ms 135.007 ms 214.243 ms 8 te-7-4-71.nycmny2wch010.wcg.Level3.net (4.68.110.22) 75.824 ms 75.695 ms 76.566 ms 9 64.200.249.153 (64.200.249.153) 282.356 ms 138.384 ms 243.104 ms 10 * * * 11 * * * 12 * * * 13 * * * 14 www.merit.edu (198.108.1.92) 112.906 ms !C 110.515 ms !C 113.418 ms !C Try Switch (swizzerland) they are testing warp tunnels - but not producting yet :) Cheers Peter and Karin -- Peter and Karin Dambier Cesidian Root - Radice Cesidiana Rimbacher Strasse 16 D-69509 Moerlenbach-Bonsweiher +49(6209)795-816 (Telekom) +49(6252)750-308 (VoIP: sipgate.de) mail: peter@peter-dambier.de mail: peter@echnaton.arl.pirates http://iason.site.voila.fr/ https://sourceforge.net/projects/iason/ http://www.cesidianroot.com/
Peter Dambier wrote:
Neal R wrote:
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then.
Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
I remember voiping over the pond, from Frankfurt, germany to New York.
We had to twist asterisk to even accept the sip. Time was between 80 and 90 msec. The experienced time was higher. Roger, Over and Out with their interstallar hamradio experience could do it, but to a normal citizen it was unuseble.
(dsl 1000 customer, close to Frankfurt)
1 krzach.peter-dambier.de (192.168.48.2) 2.918 ms 3.599 ms 3.926 ms 2 * * * 3 217.0.78.58 85.268 ms 85.301 ms 102.059 ms 4 f-ea1.F.DE.net.DTAG.DE (62.154.18.22) 102.092 ms 110.057 ms 126.310 ms 5 p2-0.core01.fra01.atlas.cogentco.com (212.20.159.38) 126.344 ms * * 6 * * * 7 p3-0.core01.ams03.atlas.cogentco.com (130.117.0.145) 132.262 ms 139.333 ms 147.174 ms 8 p12-0.core01.lon01.atlas.cogentco.com (130.117.0.198) 76.436 ms 76.444 ms 84.374 ms 9 t1-4.mpd02.lon01.atlas.cogentco.com (130.117.1.74) 99.840 ms 99.873 ms 107.508 ms 10 t3-2.mpd01.bos01.atlas.cogentco.com (130.117.0.185) 209.678 ms 217.428 ms 225.601 ms 11 t2-4.mpd01.ord01.atlas.cogentco.com (154.54.6.22) 233.514 ms * * 12 vl3491.mpd01.ord03.atlas.cogentco.com (154.54.6.210) 243.741 ms * * 13 * * * 14 ge-1-3-0x24.aa1.mich.net (198.108.23.241) 165.776 ms 174.752 ms 193.770 ms 15 www.merit.edu (198.108.1.92)(H!) 193.812 ms (H!) 201.863 ms (H!) 209.704 ms
(colo in Amsterdam)
1 205.189.71.253 (205.189.71.253) 0.227 ms 0.257 ms 0.227 ms 2 ge-5-2-234.ipcolo1.Amsterdam1.Level3.net (212.72.46.165) 0.985 ms 0.811 ms 0.856 ms 3 ae-32-54.ebr2.Amsterdam1.Level3.net (4.68.120.126) 4.235 ms 6.575 ms 1.360 ms 4 ae-2.ebr2.London1.Level3.net (4.69.132.133) 19.097 ms 12.816 ms 18.220 ms 5 ae-4.ebr1.NewYork1.Level3.net (4.69.132.109) 78.197 ms 78.769 ms 87.062 ms 6 ae-71-71.csw2.NewYork1.Level3.net (4.69.134.70) 78.068 ms 79.058 ms 89.192 ms 7 ae-22-79.car2.NewYork1.Level3.net (4.68.16.68) 142.665 ms 135.007 ms 214.243 ms 8 te-7-4-71.nycmny2wch010.wcg.Level3.net (4.68.110.22) 75.824 ms 75.695 ms 76.566 ms 9 64.200.249.153 (64.200.249.153) 282.356 ms 138.384 ms 243.104 ms 10 * * * 11 * * * 12 * * * 13 * * * 14 www.merit.edu (198.108.1.92) 112.906 ms !C 110.515 ms !C 113.418 ms !C
Try Switch (swizzerland) they are testing warp tunnels - but not producting yet :)
Cheers Peter and Karin
Hi, Over Level 3 transit from their London 2 gateway to the New York, 111 8th St. gateway: (0.0.0.0)(tos=0x0 psize=64 bitpattern=0x00) Fri Jun 29 10:56:25 2007 Keys: Help Display mode Restart statistics Order of fields quit Packets Pings Host Loss% Snt Last Avg Best Wrst StDev 1. v5-csw01.ln1.qubenet.net 0.0% 188 1.4 6.2 0.7 197.4 24.7 2. bdr01.ln1.qubenet.net 0.0% 188 1.3 5.0 0.6 214.4 26.4 3. ipcolo2.london2.level3.net 0.0% 188 1.3 1.4 0.8 2.4 0.3 4. ae-0-52.bbr2.London2.Level3.net 0.0% 188 1.4 2.8 1.0 52.2 5.9 5. ae-0-0.bbr2.NewYork1.Level3.net 0.0% 187 67.4 69.1 66.4 181.3 12.0 as-0-0.bbr1.NewYork1.Level3.net 6. ae-31-89.car1.NewYork1.Level3.net 0.0% 187 67.5 69.3 66.7 227.1 13.9 ae-21-79.car1.NewYork1.Level3.net ae-11-69.car1.NewYork1.Level3.net ae-41-99.car1.NewYork1.Level3.net 7. bdr01.ny1.qubenet.net 0.0% 187 67.4 67.4 66.8 69.8 0.4 Over a private DS-3 circuit from and to same buildings the average latency is probably 4ms less than the above. Andy.
On Thu 28 Jun 2007 (18:20 -0500), Neal R wrote:
I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link.
What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then.
Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
He'll need a Black & Dekker drill with a hammer attachment, and an absolutely prodigious stone cutting bit, a convenient wormhole, or a waiver on the laws of physics. -- Jim Segrave jes@nl.demon.net
A reasonable latency to expect between Chicago and London would be 92ms RTT. Brian Knoll -----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf Of Neal R Sent: Thursday, June 28, 2007 6:21 PM To: nanog@merit.edu Subject: trans-Atlantic latency? I have a customer with IP transport from Sprint and McLeod and fiber connectivity to Sprint in the Chicago area. The person making the decisions is not a routing guy but is very sharp overall. He is currently examining the latency on trans-Atlantic links and has fixed on the idea that he needs 40ms or less to London through whatever carrier he picks. He has spoken to someone at Cogent about a point to point link. What is a reasonable latency to see on a link of that distance? I get the impression he is shopping for something that involves dilithium crystal powered negative latency inducers, wormhole technology, or an ethernet to tachyon bridge, but its been a long time (9/14/2001, to be exact) since I've had a trans-Atlantic circuit under my care and things were different back then. Anyone care to enlighten me on what these guys can reasonably expect on such a link? My best guess is he'd like service from Colt based on the type of customer he is trying to reach, but its a big muddle and I don't get to talk to all of the players ...
participants (10)
-
Andy Ashley
-
Brian Knoll (TTNET)
-
Iljitsch van Beijnum
-
Jay Hennigan
-
Jim Segrave
-
Leigh Porter
-
Majdi S. Abbas
-
Neal R
-
Peter Dambier
-
Security Admin (NetSec)