Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

Traceroute and latency difference

Hi all,

I tried to perform traceroute from same router (A) toward 2 different sites (to B and C).

The first hop for those trace results are the same (A1), but the response time is very much different.

When tracing to site B, response by A1 is 48ms but when tracing to site C, response by A1 is 200ms.

Any idea what is going on here? both tracing are using the same VRF instance.


Re: Traceroute and latency difference

If the sites B and C wherever they connect are using different speed interconnects. (64k, E1/T1, DS3, ATM, Gig etc) then depending on that the propogation delay would be effected.

But again without knowing the exact topology, and the media used for all the connections, it difficult to say anything conclusively.



Cisco Employee

Re: Traceroute and latency difference


The ICMP TTL expired packets from a P router will have to be sent to the destination CE and back. Thus the first hop should be seen with approx. round trip time to the destination.

Imagine you traceroute to a remote CE. Up to the ingress PE/VRF everything is IP only. The time measured is the roundtrip time between the traceroute host and the answering IP router.

To send a packet across the MPLS core, two labels are needed, top label to send to the BGP next hop, bottom label is VPN label. A P router discarding this packet with labelstack, will create an ICMP "TTL expired in transit" packet. The question is, how to send it back to the originating host. As a P router does not have customer routes any direct delivery of the ICMP packet is impossible. Thus the ICMP packet is sent further on through the same LSP as the original discarded IP packet. This means, that the ICMP reply of a P router will travel to the remote CE, who does an IP lookup and sends it back. You can easily verify this with an access-list denying all ICMP packets in at the remote CE in a lab environment. This will discard the ICMP packets from the core and thus every core/P router will show up as "*" in your traceroute output.

So even the first hop appears with the complete round trip time to the remote CE.

In addition there might be variations due to overload conditions and the like.

Hope this helps! Please rate all posts.

Regards, Martin