7200emu.hacki.at

[ccie14745]

Here's the version 2.5 of my MPLS lab updated to use MPLS-TE instead of LDP for distribution of labels. There aren't any other changes over version 2, so the diagrams should be identical. Remember that this version consists of a 6-node MPLS network with a pair of P-core routers that also act as route-reflectors for the PE routers.

Hope this helps someone out there...
--
Justin White
CCIE #14745

[conspathas]

Love your work Justin! Tucking these little nuggets away for when I'm ready to hit the SP labs!

[ccie14745]

Well, take it at face value. It doesn't really use RSVP-TE for anything but label distribution. There's a lot more than can be added beyond my simple setup.

I got the idea for this from the Cisco Press book "MPLS Fundementals", which I think is a really good reference book to keep handy for all the variety of MPLS topics out there. There's a chapter devoted to Traffic Engineering and I got to thinking about what would be required to switch from LDP to RSVP-TE. I was a little surprised at what few changes were required.

Biggest thing to note is that the P routers don't have any tunnels build, but still run MPLS-TE to act as mid-points. Just the PEs have tunnels, and even then the tunnels are considered "one-way" from an LSP perspective.

Wouldn't be possible to even really discuss these things with the contributions of Greg and Chris.

[Shepardos]

Hi,

I just wonder why you use " no peer neighbor-route"? Did you realy need this?

Regards
Shepardos

[ccie14745]

It's kind of my pet peeve. PPP encapsulation will add a host route to the routing table for any remote peers. "no peer neighbor-route" simply keeps the /32 routes out of the routing table. There's instances where it's useful, like VPDN or Dial-on-Demand scenarios, but not for Serial links.

The configs will work just fine without this command. It's purely an annoyance of mine.

[Shepardos]

Did you prepare for the CCIE SP Track right now? If yes did you use the workbook from INTERNETWORKEXPERT?

[ccie14745]

While that's a thought that I've had, I am not currently preparing for that track. Actually, this push was job-related, which is a longer story that I want to detail here.

Needless to say, I would defintely be using the IEWB-SP for my preparations, if/when I decide to do so. I used their RS workbooks for my RS lab attempts and can't say enough kind things about how well I felt prepared. The Brians (as they're known) have a fantastic model for CCIE preparation and I would not have any qualms about using their materials again. I liked their presentation and arrangements much better than the other workbooks.

Of course, I'm not really intending this post to start a holy war. CCIE debates over workbooks is almost like the Ethernet vs. Token Ring debates of the old days. It gets really personal in a hurry if we aren't careful.

[eltech]

Here is the lab in GNS3 format... Definitely CPU taxing

http://www.gns3-labs.com/2008/07/10/topology-mpls-vpns-with-mpls-te-instead-of-ldp-ospf-routing-protocols/

I will attempt to create 3.0 as well ...

Excellent work Justin!