3

When running a traceroute, should I see any public IP addresses when tracing route from one MPLS point to another MPLS point? So when I do a traceroute from 10.0.0.1 to 10.0.1.1, I see some public IP address in between...

On some of the routers I see this, on others I don't, what does this mean?

3 Answers 3

3

That depend on your MPLS provider. so there are a patch for the traceroute program, that print the MPLS label using ICMP, if the LSR is configured to affich these labels ( but most of providers don't do). You may test the NANOG traceroute normalized in RFC 4950

3
  • I see MPLS Lables on the traceroute when using ping from the Cisco router, but I don't really know what they mean ... :-/ Oct 9, 2009 at 14:26
  • Errr... when using traceroute, not ping, of course Oct 9, 2009 at 14:37
  • Hi Kyle, well there are a nice tuto that you may take a look on it www.cisco.com/warp/public/105/mpls_traceroute.pdf All the process traceroute over MPLS are explained. Oct 9, 2009 at 20:51
3

Just because they are not in a reserved private IP space does not mean they are public. The private VOIP connection I have to my provider uses "public" IP addresses but there are no routes to the internet. I believe a common reason for this is to prevent addressing conflicts with customers internal ranges.

1

The IPs you see are probably the loopback (that is, "a virtual interface that is the first to become active and the last to disappear on a router, usually used for router ID and for management access") addresses of devices through the MPLS network.

I am actually a little bit surprised you see them from your router, that seems to indicate that the MPLS cloud is extended all the way to the CE (customer edge) device.

8
  • @vatine "that seems to indicate that the MPLS cloud is extended all the way to the CE (customer edge) device." I'm curious to know why it shouldn't? Could you please elaborate on that? Mar 7, 2011 at 10:02
  • @vivek It all depends on how the service provider has built the network and what service the customer needs. If all the customer needs is a "single pipe", having a customer edge device that speaks just plain IP with the provider edge means less traffic (probably not by much, but at least by one, probably two MPLS tags) and the possibility of using cheaper CE devices (as they'd only need to speak IP instead of MPLS). It also makes it possible to let the customer own/manage the CE device, something that is inadvisable if you're exposing your LDP to them.
    – Vatine
    Mar 7, 2011 at 10:20
  • @vatine so in that case say, I have two corporate offices connected over the WAN using MPLS (from a service provider) I don't really need to know anything apart from the fact that it is MPLS right? Also say I have another IPLC connection for interconnecting those two offices, is it possible to have the MPLS and IPLC in a failover setup. Mar 7, 2011 at 10:29
  • @vivek, a hesitant "maybe". You can extend an MPLS VRF over multiple VLANs, if the underlying transport supports VLAN tags. As long as you can have independent L3 routing between the sites, you can do routing-based failover. The main win between having MPLS in the SP backbone versus not having MPLS is that your IP space is isolated from that of other customers of the SP, but that is the case no matter if the MPLS terminates on the PE or the CE.
    – Vatine
    Mar 7, 2011 at 11:11
  • @vatine thanks for your time. My current scenario is that I have the option of using one MPLS and one IPLC connection for a typical H.Office and Branch office connectivity . My doubt is whether it is technically possible to use MPLS as the primary link and use IPLC as the failover (eg in a HSRP setup). Mar 7, 2011 at 11:21

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .