Re: Draft of Route-Flap Dampening Paper

[Thomas Telkamp <]

Hi Christian,

> 1.2 What is route-flap dampening ?
> 
>     If a threshold of the number of pairs of withdrawals/announcements
>     (=flap) is exceeded in a given timeframe the prefix is held down for a

>     scheduled SW upgrade of backbone router failed:
>     
>     	- reload after SW upgrade	1 flap
> 	- new SW crashed		1 flap
> 	- reload with old SW		1 flap
> 					------
> 					3 flaps within 10 minutes

You regard a withdrawal + announcement as 1 flap, right? Look at this:

 router-a#sh ip bgp 195.206.64.168 
 BGP routing table entry for 195.206.64.168/30, version 6
 Paths: (1 available, best #1)
   65001 65002
     192.168.1.1 from 192.168.1.1
       Origin IGP, valid, external, best

Now the BGP session between AS65001 and AS65002 is reset...

 router-a#sh ip bgp 195.206.64.168
 BGP routing table entry for 195.206.64.168/30, version 7
 Paths: (1 available, no best path)
   65001 65002 (history entry)
     192.168.1.1 from 192.168.1.1
       Origin IGP, external
       Dampinfo: penalty 1000, flapped 1 times in 00:00:00

...the route 195.206.64.168/30 becomes a history entry. The BGP session
comes up again...

 router-a#sh ip bgp 195.206.64.168
 BGP routing table entry for 195.206.64.168/30, version 8
 Paths: (1 available, best #1)
   65001 65002
     192.168.1.1 from 192.168.1.1
       Origin IGP, valid, external, best
       Dampinfo: penalty 1480, flapped 2 times in 00:00:32

...and the route is back, but flapped 2 times!

The reset of the BGP session could be caused by a reload of the AS65002
router and results in 2 flaps. It seems that the cisco definition of 'one
flap' is different from the one in your paper, or am I overlooking
something?  (btw, it's not clear to me why the penalty is 1480 already
after 32 seconds) 

---
Thomas