Skip to main content

OSPF Areas and Summarization may not be needed

From their design purpose, some are not true at 2019: 1

  1. Reduce CPU and memory usage is mainly because both techniques result a smaller routing table. Unless the network is huge, this is not true. The CPU power and memory are significantly cheaper than both were designed, routers can easily handle a lot of routes now. Incremental SPF has also been introduced for years as well, which reduces the calculation a lot.
  2. From the above point, the difference between using both (or one of ) methods or not is not obvious now.
  3. Isolation of impacts. This is the only benefits for use them, since a change in a area won't impact others, especially there's a link damping. We have alternatives like event dampening or automation scripts to make the link offline for a period.

The issues these techniques bring to the network are:

  1. strict hierarchical design
  2. potentially announcing the same summarized prefix to create blackholes

Some reasons may prevent us removing them, like: 2

  1. vendor specific implementation, which implies a bad smell of the code quality
  2. license issue which limits the number of routes
  3. old devices.

Footnotes​

  1. Do We Still Need OSPF Areas and Summarization? « ipSpace.net blog, Zotero ↩

  2. OSPF Areas and Summarization: Theory and Reality « ipSpace.net blog, Zotero ↩