One of the things I’m trying to develop with these practice labs is a step-by-step process that will get me through the real lab with a minimum of mistakes. At the moment, my mental process flows like this (probably similar to other published strategies that the big training companies have, no great revelations here, I know):
- Build layer 2. Turn up the frame-relay, PPP, ethernet, VLANs, trunks, q-in-q tunnels, etc. to get all links up and running. Make sure each link-end can ping itself and the other end of the link. Yes, ping itself, especially on frame links where you may have inverse arp challenges depending on the topology required in the scenario.
- Build IGP domains, one at a time. And I tweaked this part of the process quite a bit tonight. In this specific scenario, I’ve got unicast RIP with unique neighbor pairings, 2 EIGRP autonomous systems, and multi-area OSPF including NSSAs. So, I built each IGP domain out, one at a time, checking the routing tables of each router in each IGP domain to be sure that each route appeared exactly as I thought it should appear. For instance, when I completed the OSPF configuration, I went back through each OSPF router, and reviewed the OSPF routes. I checked each routing table against the network diagram, and verified every single route. Now, you might think that that was time-consuming, and indeed, it did take a little bit of time. But I feel that at this point, I HAVE to check all the routes in that manner. Why? Because one of my struggles is with redistribution. So, knowing that all my IGPs are 100% perfect before I start working through my redistribution strategy is critical. It’s a building block to make me the redistribution deity that I aspire to be.
- Establish end-to-end IPv4 reachability. This means getting the redistribution right, within the parameters of the scenario. This is a huge step, what I believe NMC calls “The Golden Moment”. You have to get this right, as a lack of reachability may make impossible other tasks required in the lab.
- Complete BGP. Depending on the requirements of the BGP section, you make have to have end-to-end reachability before you’ll be able to get all your BGP peers talking. So I layer BGP on top of a proven IGP topology.
- Build the IPv6 network. In the scenarios I’ve seen thus far, the IPv6 network is usually simpler to configure than the IPv4, with similar concerns to IPv4. Syntactically different than IPv4, IPv6 isn’t overly challenging thus far…challenging, but not killing me. My IPv4 knowledge has translated reasonably well for many of the IPv6 tasks. And while I haven’t found yet that there are other tasks dependent on the IPv6 topology being correct, I think it’s a “best practice” to get the IPv6 going, just in case I get a scenario where some of the other tasks indeed require IPv6 to be working correctly.
- Everything else. This includes multicast, security, Catalyst-specific tasks, QoS, etc. I’m finding that the tasks that fall into “everything else” can usually be done independently of the other “everything else” tasks, but often CANNOT be done unless you’ve at least made it to step 3 above, complete IPv4 reachability. In any case, my idea with leaving this step until the end is that if I’ve done pretty well with the steps above, I’ve probably got somewhere between 45 and 60 points, and then I can go after whatever “everything else” might be, in whatever order I want to keep building points well beyond the bare minimum of 80. So if I get a multicast scenario that’s worth 4 points and looks like a pain, maybe I’ll leave it until last, instead focusing on a QoS scenario that I know I can nail and is worth 6 points. If I’ve got that end-to-end reachability, I can pick and choose the rest of the tasks, right? And to my way of thinking, that will maximize my point scoring potential.