Outbound traffic from 208.89.136.0/22 going from L.A. to London?

Vincent Aniello VAniello at portware.com
Thu Sep 18 09:30:06 UTC 2014


We are seeing issues on our Savvis Internet connections in New York to 
users in London and Sweden.  Not many details yet, just seeing slow and 
sporadic connectivity.

--Vincent



From:   Todd Lyons <tlyons at ivenue.com>
To:     NANOG list <nanog at nanog.org>
Date:   09/18/2014 05:21 AM
Subject:        Outbound traffic from 208.89.136.0/22 going from L.A. to 
London?
Sent by:        "NANOG" <nanog-bounces at nanog.org>



I'm seeing some weird routing outbound for *some* destinations.
Testing from 208.89.139.252.

google.com resolves for me to 74.125.239.112 and the outbound path is
fast and appears to stay in the US.

yahoo.com resolves for me to 98.138.253.109, but the outbound path
goes through Savvis London, then to BT, then to Yahoo.

# mtr 98.138.253.109 --report --report-cycles 10
HOST: mail.mrball.net             Loss%   Snt   Last   Avg  Best  Wrst 
StDev
...internal...
  5. gw.iv.ivenue.com              0.0%    10    2.7   2.7   2.6   2.8 0.1
  6. 64.70.17.121                  0.0%    10    2.8   3.8   2.8  11.5 2.7
  7. hr2-v3007.la1.savvis.net      0.0%    10    2.8  22.9   2.8 203.1 
63.3
  8. 204.70.203.90                 0.0%    10    4.0   4.3   3.8   5.6 0.5
  9. cr1-pos-0-0-0-0.londonuk1.sa  0.0%    10  177.1 157.9 151.1 179.7 
11.5
 10. esr1-xe-7-0-0.ams.savvis.net  0.0%    10  166.5 166.4 166.3 166.5 0.1
 11. ???                          100.0    10    0.0   0.0   0.0   0.0 0.0
 12. ???                          100.0    10    0.0   0.0   0.0   0.0 0.0
 13. ixp1-xe-11-0-0-0.us-ash.eu.b 80.0%    10  597.0 595.7 594.4 597.0 1.9
 14. ???                          100.0    10    0.0   0.0   0.0   0.0 0.0
 15. ???                          100.0    10    0.0   0.0   0.0   0.0 0.0
 16. ae-5.pat1.nez.yahoo.com      90.0%    10  546.0 546.0 546.0 546.0 0.0
 17. ???                          100.0    10    0.0   0.0   0.0   0.0 0.0

# mtr 98.138.253.109 -n --report --report-cycles 15
HOST: mail.mrball.net             Loss%   Snt   Last   Avg  Best  Wrst 
StDev
...internal...
  5. 208.89.136.1                  0.0%    15    2.6   2.7   2.6   2.9 0.1
  6. 64.70.17.121                  0.0%    15    2.8   8.9   2.8  57.5 
16.2
  7. 64.70.11.57                   0.0%    15    2.8   2.9   2.8   3.2 0.1
  8. 204.70.203.90                 0.0%    15    4.4   5.4   3.7  23.5 5.0
  9. 204.70.192.122                0.0%    15  151.2 154.7 151.1 172.9 7.0
 10. 204.70.202.2                  0.0%    15  166.4 168.8 166.3 202.7 9.4
 11. ???                          100.0    15    0.0   0.0   0.0   0.0 0.0
 12. 166.49.208.90                80.0%    15  508.3 507.2 505.9 508.3 1.2
 13. 166.49.208.59                80.0%    15  589.8 589.8 588.9 590.6 0.9
 14. ???                          100.0    15    0.0   0.0   0.0   0.0 0.0
 15. 216.115.96.81                93.3%    15  521.8 521.8 521.8 521.8 0.0
 16. ???                          100.0    15    0.0   0.0   0.0   0.0 0.0
 17. 216.115.100.3                93.3%    15  537.4 537.4 537.4 537.4 0.0
 18. 98.138.144.29                91.7%    12  537.1 537.1 537.1 537.1 0.0
 19. ???                          100.0     5    0.0   0.0   0.0   0.0 0.0
 20. 98.138.240.28                80.0%     5  537.9 537.9 537.9 537.9 0.0
 21. ???                          100.0     4    0.0   0.0   0.0   0.0 0.0
 22. 98.138.253.109               66.7%     3  537.9 537.9 537.9 537.9 0.0

Traffic served by Akamai appears to be fast and stays local (i.e.
tested to www.apple.com).

A call in to Savvis got attached to master ticket 4828052, which is
for "large packet loss".  Describes the symptom, not the cause.

Is there anybody who has more info?  Anybody who's a Savvis customer
who sees the same weird outbound routing?  Or are we the only ones?
Feel free to traceroute to 208.89.139.252 to try and duplicate these
results.

...Todd

-- 
The total budget at all receivers for solving senders' problems is $0.
If you want them to accept your mail and manage it the way you want,
send it the way the spec says to. --John Levine




More information about the NANOG mailing list