This topic is to discuss the following lesson:
A few things that needs to be explained here:
Why is the penalty value getting decayed every 60 seconds, the article only mentions about one method of bringing down the penalty value i.e. half life which has a default value of 15 minutes. It is mentioned that when half life is reached the penalty is decayed by 50%
EvD: accum. penalty decayed to 526 after 60 second(s)
EvD: accum. penalty decayed to 502 after 60 second(s)
EvD: accum. penalty decayed to 479 after 60 second(s)
I am guessing that this 50% decrease is achieved by incrementally decaying the penalty every 60 seconds.
The other thing that needs explanation is:
Why route dampening is not recommended and how does it impact the route selection algorithm.
I understand your confusion. It is true that the penalty is decreased by 50% every 15 minutes. This however does not occur “all at once” when 15 minutes have elapsed. It takes place gradually over time.
From the output you have in your post, you see that after 60 seconds the penalty has been reduced by 24. It is later reduced by 23 in the next 60 seconds. The rate of reduction every 60 seconds is such that by the time 15 minutes have elapsed, the value will reach 50% of what it was 15 minutes ago.
I will ask Rene to add this detail to the lesson.
BGP dampening introduces some unwanted phenomena in BGP. RIPE NCC, or Réseaux IP Européens Network Coordination Centre is the regional internet registry for Europe, the Middle East, and parts of Asia. They’ve issued the following Working Group Recommendation concerning dampening:
In it, they mention that impeded convergence and incompatible minimum route advertisement interval on networks managed by different ISPs are problems that are induced by dampening.
Like Rene mentioned in the lesson, the CPU power of today’s network devices is more than enough to deal with BGP flapping routes. In the past, overwhelming the CPU resources of a BGP router due to flapping BGP routes was worse than the network problems introduced by the feature.
Today, it is preferable to just let BGP routes flap, rather than dampen them.
I hope this has been helpful!
Hi Rene and team.
Thank you for all you do. My question here is:
what if we still need to reach network 126.96.36.199/32 from R2 after dampening. Why not just shutdown Lo0 interface if we did not need the route on R2’s routing table?
Dampening is used to temporarily prevent changing network conditions from repeatedly installing and removing routes from BGP tables. If you still need to reach 188.8.131.52/32, well that’s life . What I mean is, there’s nothing you can do. You must simply wait until the network stabilizes, and until the penalties expire to be able to reach that network once again.
You suggest that the Lo0 interface can be shutdown to resolve the issue. Well, if you’re an admin, would you be able to determine that quickly enough to go in and shut it down? And if you do, when will you bring it back up again?
If you’re suggesting that the BGP process automatically shut down Lo0 to resolve the issue, then yes, for this topology, this would work, but what if that is an interface that connects to several other BGP routers? If you shut down that interface, you would not only lose connectivity to the network in question but to all networks reached via that interface.
Does that make sense?
I hope this has been helpful!
Thank you for clarifying that Las. Well understood.