[outages] SEA-ME-WE3 undersea cable cut

Constantine A. Murenin mureninc at gmail.com
Wed Jan 16 15:26:29 EST 2013


On 15 January 2013 21:23, Constantine A. Murenin <mureninc at gmail.com> wrote:
> On 15 January 2013 18:54, Ben Aitchison <ben at meh.net.nz> wrote:
>> On Tue, Jan 15, 2013 at 06:44:24PM -0800, Constantine A. Murenin wrote:
>>> On 15 January 2013 18:10, Ben Aitchison <ben at meh.net.nz> wrote:
>>> > Japan/Hong Kong.  But from reading about the Singapore Battle.net server for Starcraft 2, it sounded like Australia has a lot of ISP's with bad Asia routing in
>>> > general so may not make much difference.
>>>
>>> As if routing between Asia-Pacific and Europe (or even Russia) is any good! :-p
>>
>> It's actually pretty good to Korea, Singapore, Japan, and Hong Kong generally.
>>
>> It's India, Pakistan, China etc that aren't so good.
>>
>> Europe to China is terrible though.  I've got a friend who was asking about remote between UK/China.  And if it's possible to get good routes to there.
>>
>> And I can't find anyone doing good routing to China from Europe.  I'm sure there's meant to be a cable with a good path?
>>
>> Ben.
>
> Do you have a traceroute of any of these good routes?  Are they under
> 200ms or above 300ms?
>
> I've never seen a traceroute between Japan (UTC+9) and either London
> (0/+1), Germany (+1/+2) or Moscow (+4 (formerly +3/+4)) that didn't go
> through usually both NYC (-5/-4) and SJC (-8/-7).
>
> All in the northern hemisphere, timezone-wise Moscow and Japan are
> only 5 hours apart, whereas Moscow and SJC are 12 hours apart right
> now, and SJC and Japan are an extra 7 hours.  5 vs. 19? :-)
>
> Actually, there were a couple of weeks when routing between HE.net in
> FMT and a RETN.net-based connection in MSK was going through Japan one
> way, and it was adding a little extra latency.  As an end-user, that
> was the first and only time ever that I had any remote proof of any
> internet pipes leaving Russia not through Europe!
>
> http://tu.cnst.su/post/13096244490/he-net-pings-between-fremont1-and-moscow-russia
> https://plus.google.com/101080388381040783378/posts/LY6m7edSe5x
>
> But the traceroute below is representative of about the only route
> there is between Europe and Asia:
>
> # traceroute www.allbsd.org
> traceroute to www.allbsd.org (133.31.130.35), 32 hops max, 52 byte packets
>  1  [AS29182] gw.webdc.ru (188.120.247.254)  0.592 ms  0.497 ms  0.416 ms
>  2  [AS29182] xe200-40.webdc.ru (92.63.108.89)  0.422 ms  0.396 ms  0.253 ms
>  3  [AS29470] 46.46.168.173 (46.46.168.173)  0.447 ms  0.561 ms  0.718 ms
>  4  [AS9002] xe000-8.RT.TLX.NYC.US.retn.net (87.245.233.114)  124.381
> ms  124.784 ms  124.506 ms
>  5  [AS4637] NYIIX.IIJ.Net (198.32.160.42)  124.510 ms  124.533 ms  124.506 ms
>  6  [AS14936] sjc002bb01.IIJ.net (206.132.169.206)  209.920 ms
> 209.940 ms  209.929 ms
>  7  [AS14936] sjc002bf02.IIJ.net (206.132.169.249)  209.930 ms
>     [AS14936] sjc002bf00.IIJ.Net (206.132.169.241)  209.967 ms
>     [AS14936] sjc002bf02.IIJ.Net (206.132.169.249)  209.971 ms
>  8  [AS2497] tky001bf00.IIJ.Net (216.98.96.181)  314.845 ms
>     [AS14936] tky009bf01.IIJ.net (206.132.169.122)  312.017 ms
>     [AS2497] tky001bf00.IIJ.Net (216.98.96.185)  314.899 ms
>  9  [AS2497] tky009bb11.IIJ.Net (58.138.80.214)  313.616 ms
>     [AS2497] tky009bb11.IIJ.Net (58.138.80.210)  312.620 ms
>     [AS2497] tky009bb10.IIJ.Net (58.138.80.170)  313.504 ms
> 10  [AS2497] tky009ipgw11.IIJ.Net (58.138.112.158)  312.611 ms
>     [AS2497] tky009ipgw10.IIJ.Net (58.138.112.154)  314.176 ms
>     [AS2497] tky009ipgw10.IIJ.Net (58.138.112.146)  312.732 ms
> 11  [AS2497] tky009ip71.IIJ.Net (58.138.112.110)  312.879 ms
>     [AS2497] tky009ip71.IIJ.Net (58.138.112.102)  313.562 ms
>     [AS2497] tky009ip71.IIJ.Net (58.138.112.110)  313.660 ms
> 12  [AS2497] 210.138.9.166 (210.138.9.166)  322.967 ms  355.891 ms  352.757 ms
> 13  [AS55390] 133.31.14.2 (133.31.14.2)  325.548 ms  320.578 ms  316.472 ms
> 14  [AS55390] vlsi03.si.noda.tus.ac.jp (133.31.130.35)  317.491 ms
> 314.033 ms  315.634 ms
>
> The latency could have been around 130ms if it didn't go through NA.
> Wishful thinking, I know!
>
> Cheers,
> Constantine.


Actually, I digged around on http://www.cablemap.info/ and
http://www.submarinecablemap.com/ , and tried finding a looking glass
on the uber-premium Rostelecom network:


Moscow M7 to Japan in 135ms:

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=www.allbsd.org&agent=10&flt_inp=
traceroute to 133.31.130.35 (133.31.130.35), 30 hops max, 40 byte packets
 1  xe-4-0-0.110.m7-ar4.msk.ip.rostelecom.ru (87.226.136.153)  0.368
ms  0.207 ms  0.197 ms
 2  95.167.91.155 (95.167.91.155)  107.434 ms  107.419 ms  110.833 ms
 3  61.213.146.253 (61.213.146.253)  135.370 ms  139.326 ms  135.584 ms
 4  ae-12.r25.tokyjp01.jp.bb.gin.ntt.net (129.250.2.4)  151.707 ms
ae-14.r24.tokyjp01.jp.bb.gin.ntt.net (129.250.4.232)  198.260 ms
ae-12.r25.tokyjp01.jp.bb.gin.ntt.net (129.250.2.4)  151.628 ms
 5  ae-1.r00.tokyjp01.jp.bb.gin.ntt.net (129.250.5.93)  132.876 ms
136.682 ms ae-2.r00.tokyjp01.jp.bb.gin.ntt.net (129.250.5.95)  136.791
ms
 6  210.173.177.6 (210.173.177.6)  179.909 ms  179.795 ms  179.799 ms
 7  tky008bf01.IIJ.Net (58.138.80.133)  185.704 ms  189.484 ms
tky009bf01.IIJ.Net (58.138.80.141)  185.714 ms
 8  tky009bb11.IIJ.Net (58.138.80.214)  177.263 ms tky009bb10.IIJ.Net
(58.138.80.170)  185.212 ms tky009bb10.IIJ.Net (58.138.80.178)
184.635 ms
 9  tky009ipgw10.IIJ.Net (58.138.112.146)  181.265 ms
tky009ipgw11.IIJ.Net (58.138.112.158)  177.729 ms tky009ipgw11.IIJ.Net
(58.138.112.150)  178.628 ms
10  tky009ip71.IIJ.Net (58.138.112.110)  181.287 ms tky009ip71.IIJ.Net
(58.138.112.102)  178.623 ms tky009ip71.IIJ.Net (58.138.112.110)
177.705 ms
11  210.138.9.166 (210.138.9.166)  182.776 ms  186.179 ms  182.522 ms
12  133.31.14.2 (133.31.14.2)  180.277 ms  183.843 ms  183.989 ms
13  vlsi03.si.noda.tus.ac.jp (133.31.130.35)  183.191 ms  188.059 ms  183.389 ms

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=speedtest.tokyo.linode.com&agent=10&flt_inp=
traceroute to 106.187.96.148 (106.187.96.148), 30 hops max, 40 byte packets
 1  xe-4-0-0.110.m7-ar4.msk.ip.rostelecom.ru (87.226.136.153)  23.384
ms  23.265 ms  23.210 ms
 2  95.167.91.155 (95.167.91.155)  137.527 ms  107.664 ms  137.367 ms
 3  188.254.55.226 (188.254.55.226)  136.247 ms  136.225 ms  136.093 ms
 4  otejbb206.int-gw.kddi.ne.jp (106.187.6.161)  138.076 ms  138.037
ms otejbb206.int-gw.kddi.ne.jp (106.187.6.165)  133.907 ms
 5  cm-fcu203.kddnet.ad.jp (124.215.194.180)  135.735 ms  135.717 ms
cm-fcu203.kddnet.ad.jp (124.215.194.164)  147.205 ms
 6  124.215.199.122 (124.215.199.122)  136.285 ms  139.393 ms  135.604 ms
 7  speedtest.tokyo.linode.com (106.187.96.148)  138.970 ms  138.876
ms  134.921 ms


Ekaterinburg (still Europe!) to Japan in 110ms:

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=www.allbsd.org&agent=7&flt_inp=
traceroute to 133.31.130.35 (133.31.130.35), 30 hops max, 40 byte packets
 1   (79.133.95.92)  0.379 ms  0.344 ms  0.380 ms
 2   (87.226.138.209)  0.238 ms  0.254 ms  0.234 ms
 3   (95.167.95.254)  83.865 ms  87.841 ms  (95.167.91.155)  83.656 ms
 4   (61.213.146.253)  112.165 ms  112.387 ms  113.661 ms
 5   (129.250.2.4)  109.586 ms  140.520 ms  (129.250.4.232)  112.703 ms
 6   (129.250.5.93)  106.773 ms  (129.250.5.95)  115.164 ms
(129.250.5.93)  109.713 ms
 7   (210.173.177.6)  113.861 ms  117.022 ms  114.089 ms
 8   (58.138.80.125)  169.942 ms  (58.138.80.129)  155.675 ms
(58.138.80.125)  169.900 ms
 9   (58.138.80.190)  159.151 ms  (58.138.80.214)  160.605 ms
(58.138.80.190)  162.487 ms
10   (58.138.112.154)  160.683 ms  (58.138.112.146)  158.709 ms  158.762 ms
11   (58.138.112.102)  165.342 ms  (58.138.112.110)  156.283 ms  156.140 ms
12   (210.138.9.166)  148.246 ms  151.240 ms  156.397 ms
13   (133.31.14.2)  168.226 ms  164.336 ms  164.427 ms
14   (133.31.130.35)  163.525 ms  171.142 ms  166.997 ms

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=speedtest.tokyo.linode.com&agent=7&flt_inp=
traceroute to 106.187.96.148 (106.187.96.148), 30 hops max, 40 byte packets
 1   (79.133.95.92)  0.382 ms  0.334 ms  0.375 ms
 2   (87.226.138.209)  0.189 ms  0.194 ms  0.196 ms
 3   (95.167.91.155)  85.521 ms  96.442 ms  (95.167.95.254)  87.759 ms
 4   (188.254.55.226)  102.672 ms  116.064 ms  119.634 ms
 5   (106.187.6.153)  122.154 ms  (106.187.6.161)  106.985 ms
(106.187.6.165)  114.575 ms
 6   (124.215.194.164)  114.239 ms  (124.215.194.180)  116.014 ms  111.516 ms
 7   (124.215.199.122)  111.025 ms  108.996 ms  108.836 ms
 8   (106.187.96.148)  116.329 ms  112.277 ms  112.110 ms


Khabarovsk, Russia to Japan in 25ms:

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=www.allbsd.org&agent=6&flt_inp=
traceroute to 133.31.130.35 (133.31.130.35), 30 hops max, 40 byte packets
 1  188.128.5.49 (188.128.5.49)  0.811 ms  0.769 ms  0.840 ms
 2  95.167.95.254 (95.167.95.254)  1.393 ms  1.366 ms  1.381 ms
 3  61.213.146.253 (61.213.146.253)  25.869 ms  25.835 ms  25.825 ms
 4  ae-12.r25.tokyjp01.jp.bb.gin.ntt.net (129.250.2.4)  90.789 ms
90.749 ms  90.702 ms
 5  ae-1.r00.tokyjp01.jp.bb.gin.ntt.net (129.250.5.93)  26.792 ms
27.078 ms ae-2.r00.tokyjp01.jp.bb.gin.ntt.net (129.250.5.95)  28.360
ms
 6  210.173.177.6 (210.173.177.6)  27.442 ms  27.454 ms  28.003 ms
 7  tky009bf00.IIJ.Net (58.138.80.137)  75.253 ms tky001bf00.IIJ.Net
(58.138.80.121)  75.135 ms tky001bf01.IIJ.Net (58.138.80.125)  75.120
ms
 8  tky009bb11.IIJ.Net (58.138.80.206)  76.136 ms tky009bb10.IIJ.Net
(58.138.80.182)  77.810 ms tky009bb11.IIJ.Net (58.138.80.202)  75.520
ms
 9  tky009ipgw10.IIJ.Net (58.138.112.154)  95.992 ms  95.974 ms
tky009ipgw11.IIJ.Net (58.138.112.150)  105.777 ms
10  tky009ip71.IIJ.Net (58.138.112.102)  90.553 ms tky009ip71.IIJ.Net
(58.138.112.110)  90.599 ms tky009ip71.IIJ.Net (58.138.112.102)
90.321 ms
11  210.138.9.166 (210.138.9.166)  77.511 ms  76.689 ms  77.554 ms
12  133.31.14.2 (133.31.14.2)  78.953 ms  79.577 ms  80.380 ms
13  vlsi03.si.noda.tus.ac.jp (133.31.130.35)  79.222 ms  79.865 ms  79.114 ms

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=speedtest.tokyo.linode.com&agent=6&flt_inp=
traceroute to 106.187.96.148 (106.187.96.148), 30 hops max, 40 byte packets
 1  188.128.5.49 (188.128.5.49)  1.417 ms  1.337 ms  1.378 ms
 2  95.167.95.254 (95.167.95.254)  1.386 ms  1.401 ms  1.375 ms
 3  188.254.55.226 (188.254.55.226)  28.380 ms  28.371 ms  28.349 ms
 4  otejbb205.int-gw.kddi.ne.jp (106.187.6.157)  24.239 ms
otejbb206.int-gw.kddi.ne.jp (106.187.6.161)  24.137 ms
otejbb205.int-gw.kddi.ne.jp (106.187.6.157)  24.143 ms
 5  cm-fcu203.kddnet.ad.jp (124.215.194.180)  25.776 ms
cm-fcu203.kddnet.ad.jp (124.215.194.164)  25.566 ms  25.596 ms
 6  124.215.199.122 (124.215.199.122)  25.878 ms  25.486 ms  25.704 ms
 7  speedtest.tokyo.linode.com (106.187.96.148)  25.483 ms  25.503 ms  25.451 ms


(See, a lot of these routes are very asymmetrical already.)


Khabarovsk, Russia to somewhere in China in 45ms:

https://lg.ip.rt.ru/lg.cgi?query=trace&addr=www.baidu.com&agent=6&flt_inp=
traceroute to 220.181.111.147 (220.181.111.147), 30 hops max, 40 byte packets
 1  188.128.5.49 (188.128.5.49)  0.761 ms  0.664 ms  0.718 ms
 2  ge-2-0-0.uak10-ar2.dv.ip.rostelecom.ru (87.226.133.182)  1.446 ms
1.427 ms  1.382 ms
 3  188.254.55.229 (188.254.55.229)  44.369 ms  44.354 ms *
 4  202.97.58.106 (202.97.58.106)  45.114 ms  45.076 ms  45.024 ms
 5  202.97.53.225 (202.97.53.225)  45.207 ms  45.136 ms  45.188 ms
 6  202.97.53.33 (202.97.53.33)  46.681 ms  46.654 ms  46.607 ms
 7  220.181.0.42 (220.181.0.42)  48.224 ms * *
 8  220.181.16.50 (220.181.16.50)  468.688 ms  468.917 ms  469.147 ms
 9  220.181.17.146 (220.181.17.146)  46.384 ms !X * *


Sadly, no "real" networks use these routes, and Rostelecom is not
particularly known for having any settlement-free peering in Russia
whatsoever (in the old days, they had all Moscow "interconnects" go
through London).  I know of no non-premium hosting providers in Moscow
that buy transit from Rostelecom.  And no other major telecoms in
Russia seem to have any Moscow-Japan routes that don't go through NA.

Cheers,
Constantine.



More information about the Outages mailing list