NTT handing packets to Reliance (Flag Telecom) in California for BSNL block

Mailman List Mirror (Read Only)
1

Hello everyone,

I was trying to understand reason for high latency between my BSNL (AS9829)
connection and a specific Germany based server on M-Online.

I can see forward path is correct but reverse path is:

traceroute to 117.200.57.47 (117.200.57.X), 30 hops max, 60 byte packets
1 gw.giga-dns.com (91.194.90.1) [AS51167] 0.929 ms 0.915 ms 1.151 ms
2 host-93-104-204-33.customer.m-online.net (93.104.204.33) [AS8767]
1.144 ms 1.370 ms 1.608 ms
3 xe-1-1-0.rt-decix-2.m-online.net (82.135.16.102) [AS8767] 8.334 ms
8.325 ms 8.559 ms
4 xe-1-1-0.rt-decix-2.m-online.net (82.135.16.102) [AS8767] 8.043 ms
213.198.72.237 (213.198.72.237) [AS2914] 8.289 ms
xe-1-1-0.rt-decix-2.m-online.net (82.135.16.102) [AS8767] 8.023 ms
5 ae-2.r20.frnkge04.de.bb.gin.ntt.net (129.250.5.217) [AS2914] 77.508 ms
213.198.72.237 (213.198.72.237) [AS2914] 7.773 ms 7.768 ms
6 ae-1.r23.amstnl02.nl.bb.gin.ntt.net (129.250.3.179) [AS2914] 15.726 ms
xe-1-1-3.r20.frnkge04.de.bb.gin.ntt.net (129.250.5.221) [AS2914] 9.195 ms
ae-3.r22.londen03.uk.bb.gin.ntt.net (129.250.3.137) [AS2914] 43.581 ms
7 ae-1.r23.amstnl02.nl.bb.gin.ntt.net (129.250.3.179) [AS2914] 15.876 ms
xe-1-1-3.r20.frnkge04.de.bb.gin.ntt.net (129.250.5.221) [AS2914] 42.557 ms
42.534 ms
8 ae-7.r21.asbnva02.us.bb.gin.ntt.net (129.250.2.144) [AS2914] 108.990
ms 101.966 ms ae-0.r23.nycmny01.us.bb.gin.ntt.net (129.250.3.73) [AS2914]
90.236 ms
9 ae-2.r21.lsanca03.us.bb.gin.ntt.net (129.250.3.55) [AS2914] 167.662 ms
ae-1.r20.asbnva02.us.bb.gin.ntt.net (129.250.2.9) [AS2914] 131.915 ms
ae-7.r21.asbnva02.us.bb.gin.ntt.net (129.250.2.144) [AS2914] 99.925 ms
10 ae-0.r20.asbnva02.us.bb.gin.ntt.net (129.250.4.4) [AS2914] 131.452 ms
ae-2.r04.lsanca03.us.bb.gin.ntt.net (129.250.5.70) [AS2914] 162.135 ms
171.360 ms
11 ae-2.r21.lsanca03.us.bb.gin.ntt.net (129.250.3.55) [AS2914] 227.815 ms
ae-2.r04.lsanca03.us.bb.gin.ntt.net (129.250.5.70) [AS2914] 160.347 ms
ae-2.r21.lsanca03.us.bb.gin.ntt.net (129.250.3.55) [AS2914] 167.755 ms
12 ae-2.r04.lsanca03.us.bb.gin.ntt.net (129.250.5.70) [AS2914] 166.245 ms
172.954 ms 162.521 ms
13 xe-0-1-0-10.r04.lsanca03.us.ce.gin.ntt.net (198.172.90.222) [AS2914]
183.765 ms 183.748 ms 185.719 ms
14 115.249.245.201 (115.249.245.201) [AS18101] 322.389 ms 80.77.1.58
(80.77.1.58) [AS15412] 316.879 ms 218.248.255.57 (218.248.255.57) [AS9829]
315.371 ms
15 218.248.255.57 (218.248.255.57) [AS9829] 311.356 ms 311.584 ms
115.249.245.201 (115.249.245.201) [AS18101] 333.586 ms
16 218.248.173.41 (218.248.173.41) [AS9829] 359.339 ms 353.042 ms
354.021 ms
17 218.248.173.41 (218.248.173.41) [AS9829] 348.914 ms 344.156 ms
342.131 ms
18 * 218.248.173.41 (218.248.173.41) [AS9829] 354.091 ms *
19 * * *
20 * * *

I can see from Reliance's Looking glass that there direct path from EU to
India on their network. I wonder why exactly NTT handling off packets to
Reliance in California? The problem seems only for BSNL block -
117.200.48.0/20.

If destination is not BSNL and say Reliance itself...say for
Monster.co.inweb server which is in Reliance datacenter:

*Query Results:*
*Router: *Amsterdam - NL
*Command:* traceroute 220.226.205.30

*Disclaimer: Traceroute is a useful tool for determining the route a packet
takes, but it should not be used as an accurate measure of network
performance. For more information please view the Traceroute
Disclaimer<https://ssp.pme.gin.ntt.net/lg/trDisclaimer.html>
.*

traceroute to 220.226.205.30 (220.226.205.30), 30 hops max, 40 byte packets
1 as-0.r22.amstnl02.nl.bb.gin.ntt.net (129.250.2.118) 15.806 ms
0.746 ms 0.718 ms
     MPLS Label=499040 CoS=0 TTL=1 S=0
     MPLS Label=304400 CoS=0 TTL=1 S=1
2 ae-5.r23.londen03.uk.bb.gin.ntt.net (129.250.5.197) 8.180 ms
7.740 ms 8.770 ms
     MPLS Label=304400 CoS=0 TTL=1 S=1
3 ae-2.r02.londen03.uk.bb.gin.ntt.net (129.250.5.41) 16.229 ms
15.829 ms 16.326 ms
4 flagtelecom-0.r02.londen03.uk.bb.gin.ntt.net (83.231.235.238)
15.059 ms 16.032 ms 7.014 ms
5 62.216.128.106 (62.216.128.106) 8.332 ms 62.216.128.122
(62.216.128.122) 32.306 ms ge-0-0-3.0.cjr02.ldn004.flagtel.com
(62.216.128.138) 9.051 ms
     MPLS Label=412816 CoS=0 TTL=1 S=1
6 so-1-0-0.0.pjr02.ldn001.flagtel.com (85.95.25.9) 8.616 ms
so-1-0-0.0.pjr01.ldn001.flagtel.com (85.95.25.13) 19.806 ms
ge-0-1-0.0.pjr01.ldn001.flagtel.com (62.216.129.137) 8.059 ms
     MPLS Label=300480 CoS=0 TTL=1 S=1
7 so-4-0-0.0.pjr01.mmb004.flagtel.com (85.95.25.113) 280.327 ms
so-1-0-0.0.pjr02.mmb004.flagtel.com (85.95.25.138) 281.659 ms
so-3-1-0.0-pjr02.mmb004-flagtel.com (85.95.26.158) 282.889 ms
8 62.216.147.138 (62.216.147.138) 285.101 ms 695.643 ms
62.216.147.46 (62.216.147.46) 278.537 ms

Pretty much direct.

I wonder what exactly is different/wrong in case of BSNL block
117.200.48.0/20? Is BSNL selectively announcing it only from Reliance's
California based router and not any other router in Europe?

Can somehow one can test & confirm the above guess of selective
announcement?

(*Apologize if I missed some fundamental glitch error. I am new to it.*)

Thanks.

2

there's a fiber cut in that part of the world, and reliance is ... in
the middle of it (smwe-4 I think?) maybe they decided that some of
their customer traffic ought to take a less congested path?

3

Hello Christopher

Thanks for pointing out SMW4 cut as reported here -
http://www.renesys.com/blog/2012/06/smw4-break-on-south-asia.shtml

As far as I see it is likely not linked to issue. I guess it is still some
bad routing issue rather then impact of cable cut since I have seen similar
issues in past (last few months).

Thanks.

4

I have also noticed that traffic sourced in NYC destined for Qatar across NTT seems to now go from NYC -> SJC -> SNG and ends up being about 180+ms longer than just going over the atlantic.

I've seen this a few times (only with NTT routes).

Thanks,
-Drew

5

I have also noticed that traffic sourced in NYC destined for Qatar across NTT seems to now go from NYC -> SJC -> SNG and ends up being about 180+ms longer than just going over the atlantic.

  I've seen some people in the middle east/africa purchase services
in Asia based on their business needs. In the case of NTT, this
is likely the case as they are a decent sized player in Asia.

I've seen this a few times (only with NTT routes).

  See below for some further insight....

Thanks,
-Drew

From: Anurag Bhatia [mailto:me@anuragbhatia.com]
Sent: Tuesday, June 19, 2012 4:15 PM
To: NANOG Mailing List
Subject: NTT handing packets to Reliance (Flag Telecom) in California for BSNL block

Hello everyone,

I was trying to understand reason for high latency between my BSNL (AS9829) connection and a specific Germany based server on M-Online.

I can see forward path is correct but reverse path is:

traceroute to 117.200.57.47 (117.200.57.X), 30 hops max, 60 byte packets
...
13 xe-0-1-0-10.r04.lsanca03.us.ce.gin.ntt.net (198.172.90.222) [AS2914]
183.765 ms 183.748 ms 185.719 ms

  .CE.GIN.NTT.NET is customer equipment. It appears they are
preferring their customer route.

I can see from Reliance's Looking glass that there direct path from EU to India on their network. I wonder why exactly NTT handling off packets to Reliance in California? The problem seems only for BSNL block - 117.200.48.0/20.

  It is going to their customer.

If destination is not BSNL and say Reliance itself...say for Monster.co.inweb server which is in Reliance datacenter:

*Query Results:*
*Router: *Amsterdam - NL
*Command:* traceroute 220.226.205.30

*Disclaimer: Traceroute is a useful tool for determining the route a packet takes, but it should not be used as an accurate measure of network performance. For more information please view the Traceroute Disclaimer<https://ssp.pme.gin.ntt.net/lg/trDisclaimer.html&gt;
.*

traceroute to 220.226.205.30 (220.226.205.30), 30 hops max, 40 byte packets
1 as-0.r22.amstnl02.nl.bb.gin.ntt.net (129.250.2.118) 15.806 ms
0.746 ms 0.718 ms
     MPLS Label=499040 CoS=0 TTL=1 S=0
     MPLS Label=304400 CoS=0 TTL=1 S=1
2 ae-5.r23.londen03.uk.bb.gin.ntt.net (129.250.5.197) 8.180 ms
7.740 ms 8.770 ms
     MPLS Label=304400 CoS=0 TTL=1 S=1
3 ae-2.r02.londen03.uk.bb.gin.ntt.net (129.250.5.41) 16.229 ms
15.829 ms 16.326 ms
4 flagtelecom-0.r02.londen03.uk.bb.gin.ntt.net (83.231.235.238)
15.059 ms 16.032 ms 7.014 ms
5 62.216.128.106 (62.216.128.106) 8.332 ms 62.216.128.122
(62.216.128.122) 32.306 ms ge-0-0-3.0.cjr02.ldn004.flagtel.com
(62.216.128.138) 9.051 ms
     MPLS Label=412816 CoS=0 TTL=1 S=1
6 so-1-0-0.0.pjr02.ldn001.flagtel.com (85.95.25.9) 8.616 ms so-1-0-0.0.pjr01.ldn001.flagtel.com (85.95.25.13) 19.806 ms ge-0-1-0.0.pjr01.ldn001.flagtel.com (62.216.129.137) 8.059 ms
     MPLS Label=300480 CoS=0 TTL=1 S=1
7 so-4-0-0.0.pjr01.mmb004.flagtel.com (85.95.25.113) 280.327 ms so-1-0-0.0.pjr02.mmb004.flagtel.com (85.95.25.138) 281.659 ms so-3-1-0.0-pjr02.mmb004-flagtel.com (85.95.26.158) 282.889 ms
8 62.216.147.138 (62.216.147.138) 285.101 ms 695.643 ms
62.216.147.46 (62.216.147.46) 278.537 ms

Pretty much direct.

I wonder what exactly is different/wrong in case of BSNL block 117.200.48.0/20? Is BSNL selectively announcing it only from Reliance's California based router and not any other router in Europe?

Can somehow one can test & confirm the above guess of selective announcement?

(*Apologize if I missed some fundamental glitch error. I am new to it.*)

  I've seen cases where the packet goes around the world based
on the networks involved. You can usually tell due to increased
latency or looking at ping -R output.

  - Jared