[ppml] Route capacity and route churn

[Scott Leibrand]

Randy Bush wrote:
> John Paul Morrison wrote:
>   
>> Randy Bush wrote:
>>     
>>>> Not when the two biggest vendors have routers that will scale to
>>>> millions of routes
>>>>         
>>>
> <snip>
>
> in reality, they can put enough ram in these boxes, but they can not
> load churn worth squat.  it would be inappropriate for me to quote
> numbers, but let's just say "not pretty" would be an understatement.

It sounds to me like Cisco and Juniper need to work on a suitable 
replacement for route flat damping, which simply picks the more stable 
route when one of them is unstable, and slows down the rate it sends 
along updates when they come in too frequently for a specific prefix.  
 From the research presentations I've seen, implementations based on a 
few simple rules like that could significantly reduce route churn 
without reducing availability...

Is there already implementation work like this ongoing?

-Scott