[outages] Packet loss between as6453/Tata, or GBLX/nLayer

Steve Rubin ser at layer42.net
Wed Dec 21 01:03:36 EST 2011


Comcast link to 6453 full as usual I am sure.  We need an update from backdoor santa.

--
Steve Rubin <ser at layer42.net> 
Layer42 Networks

On Dec 20, 2011, at 19:13, Jeremy Chadwick <outages at jdc.parodius.com> wrote:

> I've been dealing with this for weeks now.  The asymmetric nature of the
> problem means *4* ISPs would have to get involved to really figure out
> what's causing this.  What a PITA.
> 
> * Packet loss increasing between as6453.net and Tata Communications
> * Packet loss increasing between GBLX and nLayer (at the PAIX?)
> 
> Ingress traffic (work->home) shows the first problem:
> 
> Source IP: 204.8.209.251
> Dest IP:   67.180.84.87
> 
> |---------------------------------------------------------------------------------- --------|
> |                                      WinMTR statistics                                   |
> |                       Host              -   %  | Sent | Recv | Best | Avrg | Wrst | Last |
> |------------------------------------------------|------|------|------|------|------|------|
> |             xe1-8.core2.sv1.layer42.net -    0 |  198 |  198 |    3 |    9 |   83 |    3 |
> |                  sjo-bb1-link.telia.net -    0 |  198 |  198 |    2 |    4 |   31 |    3 |
> |      tata-ic-140123-sjo-bb1.c.telia.net -   20 |  109 |   88 |    3 |    8 |   21 |    8 |
> |if-4-3254.tcore2.PDI-PaloAlto.as6453.net -    0 |  198 |  198 |    3 |    6 |   55 |    6 |
> |   if-2-2.tcore1.PDI-PaloAlto.as6453.net -    0 |  198 |  198 |    3 |    5 |   64 |    4 |
> |                           66.198.127.14 -    5 |  168 |  161 |   14 |   15 |   25 |   15 |
> |pos-1-10-0-0-cr01.sanjose.ca.ibone.comcast.net -    4 |  172 |  166 |   14 |   16 |   23 |   15 |
> |pos-1-8-0-0-cr01.sacramento.ca.ibone.comcast.net -    2 |  186 |  183 |   17 |   19 |   29 |   18 |
> |pos-0-13-0-0-ar01.oakland.ca.sfba.comcast.net -    4 |  174 |  168 |   19 |   20 |   29 |   20 |
> |te-9-8-ur04.santaclara.ca.sfba.comcast.net -    1 |  194 |  193 |   12 |   13 |  173 |   13 |
> |                           68.85.191.250 -    3 |  182 |  178 |   20 |   37 |  236 |   31 |
> |      c-67-180-84-87.hsd1.ca.comcast.net -    2 |  190 |  188 |   28 |   32 |   74 |   29 |
> |________________________________________________|______|______|______|______|______|______|
> 
> Egress traffic (home->work) shows the 2nd problem.
> 
> Source IP: 67.180.84.87
> Dest IP:   204.8.209.251
>                                                  Packets               Pings
> Host                                           Loss%   Snt   Rcv  Last   Avg  Best  Wrst
> 1. gw.home.lan                                  0.0%   192   192   0.2   0.5   0.2   1.0
> 2. c-67-180-84-1.hsd1.ca.comcast.net            0.0%   192   192  27.1  26.2  10.3 139.9
> 3. te-5-5-ur04.santaclara.ca.sfba.comcast.net   0.0%   192   192  10.3  11.9   8.0 139.2
> 4. te-0-4-0-2-ar01.oakland.ca.sfba.comcast.net  0.0%   192   192  12.1  13.2  10.2  46.3
> 5. pos-2-0-0-0-cr01.sacramento.ca.ibone.comcas  0.0%   192   192  12.6  15.3  12.0  86.6
> 6. pos-0-4-0-0-cr01.sanjose.ca.ibone.comcast.n  0.0%   192   192  17.3  19.2  15.0  69.0
> 7. pos-0-1-0-0-pe01.11greatoaks.ca.ibone.comca  0.0%   192   192  22.7  24.9  19.4 118.6
> 8. 208.178.58.1                                 0.0%   192   192  23.7  25.9  17.7  88.6
> 9. ae9-20G.scr3.SNV2.gblx.net                   0.0%   192   192  20.1  22.9  18.5 101.3
> 10. po4-20G.ar1.PAO2.gblx.net                    0.0%   192   192  51.4 116.7  18.8 1955.
> 11. xe-2-0-4.ar1.pao1.us.nlayer.net              5.2%   192   182  40.5  55.3  30.7 124.3
> 12. ae0-80g.cr1.pao1.us.nlayer.net               2.6%   192   187  38.6  54.2  28.5 128.6
> 13. as8121.ae0-3001.cr1.pao1.us.nlayer.net       1.0%   192   190  70.6  36.8  26.7 105.8
> 14. xe1-4.core1.svk.layer42.net                  2.1%   192   188  27.6  33.3  27.2  86.9
> 15. ???
> 
> This is not ICMP prioritisation -- I feel this latency and packet loss
> when using Remote Desktop and SSH.  It's really quite annoying.
> 
> Obviously the asymmetric nature means either #1 or #2 could explain the
> other.  Issue smells of a capacity problem (e.g. bottleneck), but heck
> if I know. 
> 
> I look forward to insights, and/or comments from anyone at any of these
> companies, including off-list if need be.
> 
> -- 
> | Jeremy Chadwick                                jdc at parodius.com |
> | Parodius Networking                       http://www.parodius.com/ |
> | UNIX Systems Administrator                   Mountain View, CA, US |
> | Making life hard for others since 1977.               PGP 4BD6C0CB |
> 
> _______________________________________________
> Outages mailing list
> Outages at outages.org
> https://puck.nether.net/mailman/listinfo/outages




More information about the Outages mailing list