Unusual routing to Cambridge
17-07-2006, 14:13
|
#1
|
|
cf.geek
Join Date: Oct 2003
Location: Cambridge
Posts: 567
|
Unusual routing to Cambridge
This is a traceroute from my place of work to my home cable modem in Cambridge:
Tracing route to [82.6.105.xxx]
over a maximum of 30 hops:
1 <1 ms <1 ms <1 ms route-cent-3.cam.ac.uk [131.111.141.62]
2 <1 ms <1 ms <1 ms route-enet-3.cam.ac.uk [192.153.213.193]
3 <1 ms <1 ms <1 ms cambridge-bar.ja.net [146.97.40.49]
4 3 ms 3 ms 3 ms po11-0.lond-scr.ja.net [146.97.35.9]
5 3 ms 3 ms 3 ms po0-0.lond-scr4.ja.net [146.97.33.34]
6 3 ms 3 ms 3 ms po0-0.uk-gw1.ja.net [146.97.35.234]
7 3 ms 3 ms 3 ms tele-ic-1-ge-100-207.inet.ntl.com [212.250.14.33]
8 6 ms 6 ms 6 ms nth-bb-b-so-600-0.inet.ntl.com [62.253.184.1]
9 6 ms 6 ms 6 ms nth-bb-a-so-400-0.inet.ntl.com [213.105.172.193]
10 10 ms 9 ms 9 ms gfd-bb-b-so-400-0.inet.ntl.com [62.253.185.98]
11 11 ms 11 ms 11 ms bre-bb-a-so-010-0.inet.ntl.com [213.105.172.149]
12 11 ms 11 ms 11 ms bre-bb-b-ge-000-0.inet.ntl.com [213.105.174.226]
13 22 ms 22 ms 22 ms ren-bb-a-so-100-0.inet.ntl.com [62.253.185.165]
14 19 ms 19 ms 19 ms lee-bb-b-so-600-0.inet.ntl.com [62.253.185.162]
15 28 ms 28 ms 28 ms ren-bb-a-so-000-0.inet.ntl.com [62.253.185.161]
16 34 ms 28 ms 28 ms bre-bb-b-so-110-0.inet.ntl.com [62.253.185.166]
17 28 ms 28 ms 28 ms bre-bb-a-ge-000-0.inet.ntl.com [213.105.174.225]
18 59 ms 50 ms 29 ms win-bb-b-so-600-0.inet.ntl.com [62.253.185.198]
19 29 ms 29 ms 29 ms pop-bb-a-so-010-0.inet.ntl.com [62.253.185.201]
20 29 ms 29 ms 29 ms cmbg-t2core-a-pos31.inet.ntl.com [62.253.188.194]
21 29 ms 29 ms 29 ms cmbg-t2cam1-a-ge-wan31.inet.ntl.com [80.1.201.26]
22 29 ms 29 ms 29 ms cmbg-cmbg-ubr-8-ge01.inet.ntl.com [80.1.202.78]
23 38 ms 36 ms 54 ms cpc1-cmbg8-0-0-custZZZ.cmbg.cable.ntl.com [82.6.105.xxx]
So let's just check that out:
From Telehouse to Northampton.
Then to Guildford.
Then to Brentford.
Then to Renfrew, Glasgow.
Then to Leeds.
Back to Renfrew again!
Back to Brentford again!
Then to Winnersh.
Then to Poplar.
Finally, to Cambridge.
Might this, one wonders, be capable of some optimisation?
The routing in the other direction is much more conventional:
Tracing route to [128.232.xxx.yyy]
over a maximum of 30 hops:
1 3 ms <1 ms <1 ms linksys [192.168.1.1]
2 8 ms 16 ms 8 ms cmbg-ubr8 [10.128.196.1]
3 10 ms 10 ms 7 ms cmbg-t2cam1-a-ge913.inet.ntl.com [80.1.202.77]
4 10 ms 9 ms 8 ms cmbg-t2core-a-ge-wan62.inet.ntl.com [80.1.201.37]
5 15 ms 24 ms 17 ms cmbg-t2core-b-ge-wan64.inet.ntl.com [80.1.201.2]
6 15 ms 12 ms 21 ms nth-bb-b-so-210-0.inet.ntl.com [62.253.188.197]
7 19 ms 17 ms 16 ms tele-ic-1-so-700-0.inet.ntl.com [62.253.184.2]
8 16 ms 15 ms 46 ms UK-GW1.JA.NET [212.250.14.34]
9 45 ms 19 ms 15 ms po3-3.lond-scr4.ja.net [146.97.35.237]
10 15 ms 16 ms 15 ms po0-0.lond-scr.ja.net [146.97.33.33]
11 19 ms 19 ms 35 ms po0-0.cambridge-bar.ja.net [146.97.35.10]
12 19 ms 21 ms 20 ms route-enet-3.cam.ac.uk [146.97.40.50]
13 36 ms 20 ms 37 ms route-cent-3.cam.ac.uk [192.153.213.194]
14 37 ms 41 ms 18 ms zzzz.cam.ac.uk [128.232.xxx.yyy]
But of course, the round-route times are bad because of the reverse routing.
|
|
|
17-07-2006, 14:27
|
#2
|
|
"Why I oughta..."
Join Date: Jun 2006
Location: Cambridge
Services: Sky TV, VM TV, 2meg bb, radio, tel, and a lobster (but the lobster died).
Posts: 3,149
|
Re: Unusual routing to Cambridge
Very interesting!
I just tried a tracert from work (London) to home (Cambridge), but get "destination net unreachable". However, my web server at home is switched on and working. Is it something to do with my use of DynDNS?
|
|
|
17-07-2006, 17:59
|
#3
|
|
Busy Admin
Join Date: Oct 2003
Location: Nottingham
Age: 45
Services: ntl Phone : Sky+ (with multiroom) : ntl Cable (20 Mbps)
Posts: 14,423
|
Re: Unusual routing to Cambridge
Maybe it's just temporary due to a fault somewhere ?
__________________
Click here for a real, interactive, tv guide.
|
|
|
18-07-2006, 00:09
|
#4
|
|
Permanently Banned
Join Date: Apr 2006
Posts: 562
|
Re: Unusual routing to Cambridge
Robin, the ntl core appears to do some weird things on the way through as it's using static LSPs for routing decisions, check this out:
cthomas@newcore[~]$ traceroute cmbg-cmbg-ubr-8.inet.ntl.com
traceroute to cmbg-cmbg-ubr-8.inet.ntl.com (82.7.202.22), 64 hops max, 44 byte packets
4 tele-ic-1-ge-200-x.inet.ntl.com (212.250.14.x) 1.369 ms 1.133 ms 1.218 ms
5 nth-bb-b-so-110-0.inet.ntl.com (62.253.184.1) 3.728 ms 3.852 ms 3.584 ms
6 lee-bb-a-so-600-0.inet.ntl.com (62.253.185.102) 55.324 ms 6.794 ms 6.923 ms
7 nth-bb-b-so-100-0.inet.ntl.com (62.253.185.101) 6.766 ms 6.720 ms 6.605 ms
8 cmbg-t2core-b-pos31.inet.ntl.com (62.253.188.198) 8.191 ms 8.314 ms 8.402 ms
9 cmbg-t2cam1-b-ge-wan31.inet.ntl.com (80.1.201.154) 8.134 ms 8.219 ms 14.226 ms
10 cmbg-cmbg-ubr-8-ge02.inet.ntl.com (80.1.202.206) 8.987 ms * 8.246 ms
Note it visits the same router twice! (nth-bb-b)
Tracing to the *interface* on the uBR rather than its' loopback:
cthomas@newcore[~]$ traceroute 80.1.202.206
traceroute to 80.1.202.206 (80.1.202.206), 64 hops max, 44 byte packets
4 tele-ic-1-ge-200-x.inet.ntl.com (212.250.14.x) 1.424 ms 1.007 ms 1.426 ms
5 nth-bb-b-so-110-0.inet.ntl.com (62.253.184.1) 3.564 ms 5.593 ms 4.615 ms
6 cmbg-t2core-b-pos31.inet.ntl.com (62.253.188.198) 4.753 ms 5.545 ms 4.880 ms
7 cmbg-t2cam1-b-ge-wan32.inet.ntl.com (80.1.201.166) 4.791 ms 4.753 ms 5.017 ms
8 cmbg-cmbg-ubr-8-ge02.inet.ntl.com (80.1.202.206) 5.550 ms * 5.099 ms
Clearly this is policy based routing, the uBR's loopback would be received at t2cam via an IGP and be subject to different policy via route maps from the directly connected interface 80.1.202.206.
Having spoken to a member of staff a while ago he informed me that the LSPs across ntl's MPLS core are static. This would be supported by the first traceroute as it's impossible for MBGP to route traffic to the same router twice in a single trace - clearly policy has been defined to force traffic down that path.
I find it as ridiculous as you do and can only think there's a reason for forcing traffic through Leeds, perhaps the sampled on the Leeds router is getting some exercise? I suspect it's being done with static LSPs and probably RSVP is being used somewhere in the mix.
The above Paul M isn't a fault, traffic has been taking that path via nth-bb-b twice with Leeds in between for several months. One can only conclude it's intentional.
EDIT: Sorry to those who read the above and think 'wtf?!', I hope it makes some sort of sense, routing is being done based on a policy that's been defined not on what the optimal and most efficient routes are.
Last edited by James Henry; 18-07-2006 at 00:14.
|
|
|
18-07-2006, 00:23
|
#5
|
|
stringy
Join Date: Jun 2003
Location: Cleethorpes
Age: 55
Services: VM XLplus
Posts: 15,314
|
Re: Unusual routing to Cambridge
does the nottingham caching remaining disabled in my part of Grimsby/Cleethorpes make any difference, it is very unusual
__________________
Gaz
|
|
|
18-07-2006, 01:11
|
#6
|
|
cf.geek
Join Date: Jan 2004
Posts: 897
|
Re: Unusual routing to Cambridge
Quote:
|
Originally Posted by James Henry
5 nth-bb-b-so-110-0.inet.ntl.com (62.253.184.1) 3.728 ms 3.852 ms 3.584 ms
7 nth-bb-b-so-100-0.inet.ntl.com (62.253.185.101) 6.766 ms 6.720 ms 6.605 ms
Note it visits the same router twice! (nth-bb-b)
|
assuming you meant those two hops, the hosts and ips are both different so how can you say its the same router?
|
|
|
18-07-2006, 09:05
|
#7
|
|
Permanently Banned
Join Date: Apr 2006
Posts: 562
|
Re: Unusual routing to Cambridge
Quote:
|
Originally Posted by dev
assuming you meant those two hops, the hosts and ips are both different so how can you say its the same router?
|
Umm they are both nth-bb-b - Northampton backbone B, just two different interfaces:
5 nth-bb-b-so-110-0.inet.ntl.com = Northampton backbone b, serial 1/1/0.0
7 nth-bb-b-so-100-0.inet.ntl.com = Northampton backbone b, serial 1/0/0.0
I could even hazard a guess that both of the above are STM16s, 2488.32 Mbps as they share the same FPC.
Different interfaces on routers have different IP addresses ya know...
EDIT: To be even more silly Serial 1/1/0.0 is Flexible PIC Concentrator 1 / Physical Interface Card 1 / Small Form-factor Pluggable 0 . Unit 0. The DNS tells you where the interface is on the router - how cool is that?
Last edited by James Henry; 18-07-2006 at 09:09.
|
|
|
18-07-2006, 14:17
|
#8
|
|
cf.mega poster
Join Date: Nov 2003
Location: Grimsby
Posts: 1,995
|
Re: Unusual routing to Cambridge
ded kool
|
|
|
18-07-2006, 14:18
|
#9
|
|
Permanently Banned
Join Date: Apr 2006
Posts: 562
|
Re: Unusual routing to Cambridge
Quote:
|
Originally Posted by jtwn
ded kool
|
Cheers for that. If/when you come to troubleshoot large networks you'll appreciate it a lot more
|
|
|
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
|
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT +1. The time now is 01:03.
|