[arin-ppml] Routing table growth, was Re: IPv4 is depleted today

[Stephen Sprunk]

Randy Bush wrote:
>> 10. This model works "successfully" (although without any real
>>     of hierarchical aggregation due to extensive PI reuse) and
>>     provides us a small number of additional year of growth
>>     before collapsing due to departure from RFC 2008/BCP 7.
>>     
>
> while i am not claiming the bgp sky is not falling, we actually have no good measurements or experiments to substantiate the arguments that we have a particularly low ceiling.
>
> my guess is that, with current routers, it is the multi-homed enterprise edge that is likely to hit any wall before the big isps.  the latter may just see an increase in the rate of needing to appease their line card addiction.  wuzza wuzza.
>   

However, any leaf network (enterprise or otherwise) that is feeling the 
pinch can add a default (if they don't already have one) and drop 
however many routes are necessary to get down to their hardware's 
capacity.  IMHO, the real pinch is going to happen in the DFZ, where 
that isn't an option, and the most logical approach is that folks with 
long prefixes are going to find their reachability gradually decreasing 
over time.

As it stands, it seems most folks aren't filtering even at RIR minima 
yet.  That is the first step and a relatively painless one (Got a 
problem?  Quit announcing your /20 as 16 /24s!).  However, the step 
after that is be necessity going to be really painful, and the Swamp is 
the most obvious and annoying target...

S