Now before I get into this, I have to comment about a few things. This is a long-ish chapter, around 50 pages of heavy content. And it’s crammed wall-to-wall with information. There’s no slack pages here. There are a few pages of IOS output more towards the end of the chapter, but still, there’s a lot here. For all that, I think the author cheated OSPF a bit. It’s hard to criticize because they’re trying to cram a vast amount of knowledge into a single book for the CCIE R&S written exam. Even so, this chapter just isn’t well laid out. Compared to the previous chapters, I felt this chapter on OSPF was written by someone in a hurry who wanted to get all the important stuff in, but not really explain them as much as was necessary. That said, everything’s there that we’ll likely need to know. This is the “official” guide after all right? :-) Moving on…I’ll do my best to summarize this chapter as I have the previous. I know I’m not great, but I’m trying to put together a “cliff notes” version of a book that’s already a pretty well condensed tome.
OSPF stands for “open shortest path first”. OSPF is a link-state routing protocol, meaning that an OSPF router receives information from other routers about their ability to access a particular network. OSPF routers don’t exchange routes per se, but rather link states – the link a router may have to a particular network. An OSPF router takes all link state advertisements (LSA’s) from his OSPF neighbors, compiles them into his link state database (LSDB), runs the shortest path first (SPF) or Dijkstra algorithm to determine the most efficient way to forward traffic destined for a particular subnet. That SPF-calculated route is what makes it into the IP routing table of the router. So then, a significant part of understanding OSPF revolves around understanding the different types of LSA’s and their purposes.
At least, that’s how the book lays it out. But there’s a couple of other OSPF elements that one needs to understand before making much sense of the OSPF LSA’s and LSDB.
- A large OSPF network is made up of logical areas, with a central backbone area of 0, to which every other area must directly connect. If you have areas 0, 1 and 2, traffic cannot traverse from area 1 to area 2 directly. Rather, traffic will traverse from area 1 to area 0, then area 0 to area 2.
- Routers that border 2 or more areas are called area border routers, or ABR’s. ABR’s advertise different types of LSA’s into non-backbone areas.
- Routers that border an OSPF area and a non-OSPF area (i.e. some other part of the network not governed by OSPF) are called autonomous system boundary routers (ASBR’s), and they too have different types of LSA’s they advertise into the OSPF network.
- OSPF will always prefer an intra-area route (stay inside the area) over and inter-area route (hop through area 0)…even if the intra-area route traverses a woefully slow link that would have been much faster traversing via area 0. This is a major consideration for anyone doing enterprise OSPF design, especially in a WAN environment of high-cost, low-bandwidth links.
More in the next article…coffee break.