Saturday, August 13, 2011

Pros and Cons of OSPF

OSPF is by far the most deployed IGP in enterprise networks today (aside from proprietary IGPs). Some folks don’t know that there are a couple of versions of OSPF. When engineers talk about OSPF they are actually referring to OSPF version 2, or OSPFv2. The original RFC for OSPFv1 is RFC1131 published in October 1989. RFC1247 updated OSPF to version 2 in July 1991 and the current RFC is RFC2328. There have been small updates to OSPF to keep up with changing network trends – like traffic engineering extensions (RFC4203). OSPF has proven to be adaptable and flexible over the years.

So lets get down to the brass tacks.

What are the pros of OSPF?

The primary benefit to OSPF is it ubiquitous use in the Enterprise network today. Almost every trained network engineer, CCNA, JNCIA, ETC, have had some exposure to basic theory and operation of OSPF. Most people understand the concept of areas and how an area help the network scale. OSPF is comfortable to most engineers.

Speaking of scale, OSPF scales very well in almost every network out there today. The availability and use of different types of areas (normal, stubby, not-so-stubby, and totally stubby/nssa) allows you to manipulate to amount of prefixes in specific parts of the network. Only ISIS gives similar options. When network segmentation is required, OSPF offers flexibility. On the flip side, most networks today only require a single OSPF area.

Another pro for OSPF is it’s ability to support traffic engineering extensions. If you think you are going to run MPLS in your network and want to do traffic engineering – you will need a IGP that supports traffic engineering extensions. While not turned on by default (like ISIS) a simple command under your OSPF configuration usually remedies that problem. In OSPF, traffic engineering allows your LSPs to support features such as fast reroute, node & link protection, and standby links.

What are the cons of OSPF?

The first “con” that comes to mind is that OSPFv2 does not support multiple protocols. OSPFv2 supports IPv4. You need a IGP for IPv6 – you need something else like OSPFv3. They work the same, but that is still two protocols to support for “IP”. (Well, they may work the same – but Cisco for example implemented them differently in the IOS CLI.)

Another con is the backbone area (area 0). The backbone area is always designated as the number 0. You must be sure not to segment the backbone area. Sometimes this is easy and other times, based on the topology, it can be difficult. In the event of a merger or acquisition, you may have to redesign your OSPF deployment to deal with two area 0′s (<- bad).

1 comment: