0

Let's say, that there is an enterprise network with four IP transit providers:

BGP setup

As seen above, two of those IP transit providers connect to routers in one data center and other two connect to routers in another data center. Enterprise network 4.4.0.0/16 allocation is divided into 4.4.0.0/17(used in DC 1) and 4.4.128.0/17(used in DC 2).

The enterprise announces more specific prefixes to Internet in addition to 4.4.0.0/16 network in order to avoid an asymmetric routing and a situation where some of the return traffic from the Internet would be blackholed in case the connectivity between the data centers is down and a request was sent from DC 1 while the response came in from DC 2 or the other way around.

On the other hand, the reason for announcing the 4.4.0.0/16 in addition to two /17 networks is to ensure that the connectivity works even when two ISPs in the same data center are down assuming that the connectivity between the data centers is up.

Is there a better way to accomplish the same level of redundancy without polluting the Internet route table with longer prefixes?

1
  • It's "polluted" well beyond fixing already, so do whatever you want. (many would just announce /24's to minimize the possibility of a hijack)
    – Ricky
    Mar 8 at 21:42

1 Answer 1

3

As of yesterday (March 7, 2024) there are 965,225 prefixes in the "no default" zone. I don't think adding one more is going to cause much "pollution."

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.