cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
805
Views
0
Helpful
8
Replies

BGP Routing

worldcalltel
Level 1
Level 1

Hi, we are having some high latency issues on our network from NY to Philippines. I've conducted a trace route from NY to Philippines (see below). My question is, based the logs is it correct if I conclude that there is a loop going on ? or was there any sign of bgp routing loop based the logs ?

traceroute 210.5.77.146

Type escape sequence to abort.

Tracing the route to 210.5.77.146

1 4.78.179.21 [AS 3356] 0 msec 0 msec 0 msec - US; Bridgewater

2 4.68.97.62 [AS 3356] 0 msec 4 msec 0 msec - US; Kansas

3 4.68.16.75 [AS 3356] 0 msec 0 msec 0 msec - US; Kansas

4 216.6.97.21 [AS 6453] 12 msec 0 msec 0 msec - Canada; Montreal

5 216.6.97.5 [AS 6453] 0 msec - Canada; Montreal

216.6.82.93 [AS 6453] 0 msec 0 msec - Canada; Montreal

6 216.6.57.41 [AS 6453] 0 msec - Canada; Montreal

66.110.14.21 [AS 6453] [MPLS: Label 1007 Exp 0] 20 msec 20 msec - Canada; Montreal

7 66.110.27.6 [AS 6453] [MPLS: Label 2957 Exp 0] 72 msec 72 msec - Canada; Montreal

209.58.124.26 [AS 6453] [MPLS: Label 3177 Exp 0] 72 msec - Canada; Montreal

8 216.6.29.25 [AS 6453] [MPLS: Label 2278 Exp 0] 72 msec 72 msec - Canada; Montreal

216.6.95.93 [AS 6453] 248 msec - Canada; Montreal

9 216.6.29.66 [AS 6453] [MPLS: Label 3177 Exp 0] 72 msec 72 msec - Canada; Montreal

116.0.82.38 [AS 6453] 308 msec - China

10 58.71.0.131 [AS 9299] 548 msec - Philippines; Baguio

216.6.95.117 [AS 6453] 248 msec - Canada; Montreal

58.71.0.131 [AS 9299] 312 msec - Philippines; Baguio

11 210.14.3.238 [AS 9299] 312 msec - Philippines; Baguio

116.0.82.38 [AS 6453] 548 msec - China

210.14.3.238 [AS 9299] 312 msec - Philippines; Baguio

12 210.5.77.146 [AS 9299] 312 msec 312 msec * - Philippines; Makati

8 Replies 8

Danilo Dy
VIP Alumni
VIP Alumni

EDIT: The multiple IP Address in hop 5-6-7-8-9-10-11, shows that there are multiple link/route to the next hop.

I would like to know, what is the possible answer to this issue ? I already escalated the problem to our carrier. And I think they already did the re-routing. What trigger this behavior on the routing ? Thanks

The AS6453 re-appearing after hop 10 (AS9229) is strange. This suggest poor BGP configuration.

Thanks.

Hello Jaime,

what you see can be the sign that the two AS 6453 (teleglobe, ARIN) and 9299

AS9299

as-name: IPG-AS-AP

descr: Philippine Long Distance Telephone Company

descr: Philippine Internet Exchange

registered with APNIC

These two can have more the one peering arrangment in two different locations

This is not necessary a sign of a problem.

Hope to help

Giuseppe

Mohamed Sobair
Level 7
Level 7

Hi,

Possible routing loop occurs at hop 10 as it clearly shows in the trace:

10 58.71.0.131 [AS 9299], AS 9299 is taking AS 6453 as a valid path to Network 210.5.77.146 which sholudnt be.

HTH

Mohamed

Harold Ritter
Cisco Employee
Cisco Employee

Jaime,

This output doesn't show a routing loop but rather that the router at step 10 has two paths to destination, one via AS6453 and one via 9299, hence step 11 showing 6453 again.

Regards

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

Hello,

I don't know if you managed to resolve this issue already. In any case, I agree with previous posts suggesting that this traceroute output doesn't document a routing loop or any other apparent issue. If you recall how traceroute works, foreach TTL (starting from TTL=1 up to destination) there are 3 different packets sent by default. The trace doesn't show all the hops as traversed by any single of those packets (it only shows part of the story). The first indication of multiple possible paths occurs from step 4 to step 5 (since there are 2 different addresses at step 5). At step 10 you might be seeing the cumulative effect of possible paths from step 4 to step 5 for packets sent with TTL=10.

In order to troubleshoot such issues, you should have a baseline of what is (approximately, because paths change from time to time) the expected traceroute output from your network under normal circumstances and what is the expected delay as it is observed from trace/ping. I do not know what your delay used to be, but the delay approximated by the particular traceroute at the end doesn't seem excessive to me for a high distance international path (the delay reported from intermediate hops is not a certain thing to rely on, especially if there are mpls nodes there).

Still, note that when international circuits fail, they can be rerouted in a backup path that can have somewhat increased delay so you might want to check if any additional delay observed is increasing or is constant (to give you an indication: from eastern Europe to NY main path can have ~120ms RTT, while a backup path can have ~200ms RTT). In such cases, a backup is better than nothing, but you can still ask if there are any such issues (e.g. in the link from Canada to China) and when they will be resolved.

In general, when you communicate with your provider you should mention the source address of your ping/traceroute. A real problem might well be there, but manifest itself only from particular source addresses, so you might need to perform various tests (e.g. when you trace with source address in the subnet of the interconnection link with your provider everything is fine, while from any other of your addresses there are issues). This will help them trace back to you and they might see more (your trace can't show the path of the return traffic).

Kind Regards,

M.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: