MPLS VPNs with RIPv2 Complete

Wow, I actually am getting better at this MPLS stuff. When moving from static routes to RIP, the complexity increases greatly. It's a good thing RIP itself is such a simple protocol.

So the main difference here is that address-families must be created under global RIP process. All the configuration is then applied under each RIP address-family. I found it very interesting at first that none of the redistribution in BGP or RIP requires the naming of the address-family. The router is smart enough to keep track of all this based on the address-family it is under, and the vrf assigned to the interface.

For example, when I first configured RIP, I was concerned about the network statement. Being that multiple interfaces fell under 10.0.0.0, I thought all of the interfaces would be advertised into the particular address-family of RIP and that I'd have to use passive-interface or filtering to block them. Nope. The router only advertises the interfaces that have the vrf of the applicable address-family. Makes perfect sense, and makes my life a lot easier.

I also got stuck for a bit because I couldn't find the RIP routes getting distributed into BGP. Turns out I needed to use sh ip bgp vpnv4 vrf instead of sh ip bgp. Makes sense, as I needed to see the routing table of the vrf instance, not the global bgp table.

I'm feeling a lot better about MPLS VPNs today, and have a better understanding the significance of route descriptors as well.

MPLS VPNs with Static Routing Complete

Points to remember when configuring MPLS VPNs:

First create the vrf and associate the rd and import/export rules
Next assign the interfaces to the vrf and reapply IP addresses
Then configure the BGP vpnv4 address family, activating the other PE router and send both communities
Also under BGP, configure the ipv4 vrf address family, redistributing as appropriate
Finally, configure any necessary ip route vrf's.

Use show ip route vrf *, show ip bgp vpnv4 all, and ping vrf to verify connectivity

I really need to commit to memory when to use the vpnv4 address family and when to use the ipv4 vrf address family. It makes sense conceptually. The vpnv4 family includes the extended communities, and basically acts as a type of tunnel between PE routers. Then, each ipv4 vrf is like a pvc inside of that tunnel. Finally, the vrf itself describes what routes get leaked between each individual vrf.

The router uses rd's to keep track of vrf's. We use the vrf name to associate the vrf with interfaces and bgp.

The router uses rt's to advertise vrf's across vpnv4 through the use of extended communities.

SP Volume I Basic MPLS Complete

The biggest challenge here was in getting Dynamips up and running and configured. Fortunately, IE has a great guide which saved me a lot of work. I first tried to run it in Windows, but the performance was too poor. So I switched to Linux, and I can get the Vol I equipment up and running just fine.

The Basic MPLS lab was just what it says. Just needed to throw OSPF and BGP on a few routers and then put MPLS on the core. It was a nice exercise just to watch the pings propagate even though the core routers don't have a route to the final destination.

I had two mistakes in my configuration that I was able to correct pretty quicly. First, I missed mpls ip on one of the interfaces. Because it was missing, some of my hops were untagged in the mpls forwarding table. Simply checking the mpls neighbors at each router pointed me to the issue pretty quickly.

The second mistake was that I didn't put next-hop-self on the PE routers. This is necessary because the core needs a route to the next BGP hop, just not to the final destination. By deafult, iBGP does not update the next-hop and leaves it as the first eBGP hop.

My New MPLS Strategy

Now the other 3 CCIP exams are out of the way and it's time to focus on MPLS. I usually pick up networking technologies pretty easily, but there are a few in the past I have had a hard time with. One of those, in particular, was BGP.

This is one of my favorite responses to one of my groupstudy posts when I was struggling with BGP and another member made a humerous analogy to learning BGP from a book.

I have to agree with that to a certain extent. I was completely clueless on BGP a few years ago, eventhough I read Internet Routing Architectures cover to cover at least twice. It wasn't until I performed enough complex redistribution labs and worked with it in the real world before I got my arms around it.

I now seem to be having the same struggles with MPLS. I have the fundamentals more or less memorized, but I don't really understand the how's and why's of what's going on. I think that is what's keeping me from getting through the MPLS exam.

So rather than continue to brute force the exam, I've decided to pick up the InternetworkExpert Service Provider material. I'm hopeful reviewing their Class on Demand and working through their techonology labs will help to get me up to speed.

A Little Redemption

I passed the BGP exam today with a lot of room to spare. There were a lot more questions on this than on the BGP+MPLS exam, with a lower passing score to boot. A few of the questions were tricky, but for the most part they were pretty much what I was expecting.

So on the bright side, I now know I was doing pretty well on the BGP portion of the BGP+MPLS exam. The problem with this is I was doing a heck of a lot worse on the MPLS portion. If my scores were similar, that would mean I must have been getting approximately 50% on the MPLS portion of the combined exam.

Now it's time to focus on MPLS. I've tried just reading books and that doesn't seem to help it sink in. I ran into the same thing with BGP a few years back and just didn't "get it" until I did enough labs. It may be time for the same approach with MPLS.

The (Alleged) Recession's Impact on CCIE Numbers

With the performance of the stock market and layoff announcements rising, even though it's not official yet I think it's pretty safe to say the US has been in a recession for at least the past quarter.

The question I ask is, does a recession make more or less people attain their CCIE? I could see two possibilities here. First, perhaps the fear of a tightening job market would drive some to get this certification knocked out. On the other hand, having less money to pay for training and lab attempts, along with perhaps a lower possibility of rewards from passing, may cause some to put the CCIE on the back burner.

Between March and September, the rate of attainment of new CCIEs was 9.7 per day. According to InternetworkExpert.com, their latest CCIE number as of November 5th was 22542, vs 21910 on September 3rd. Therefore, over the past two months, this equates to 10.0 CCIE numbers awarded per day.

Now this is hardly a scientific study, since I don't have access to the exact amount of CCIEs given out on a particular day and I'm comparing a 6 month timeframe to a 2 month timeframe.

But it certainly does appear that this global economic slump is having no effect, or even a slightly positive effect, to the attainment of new CCIEs.

BGP Exam 642-661

I'm gonna take a stab at this exam sometime this week, possibly as early as today. From looking at the exam topics I just couldn't imagine having too much difficulty with this. Aside from ORF and Hierarchical Route Reflectors, I have quite a bit of experience on the rest of items.

I hope I can pass this today and get myself back on the right track.