Re: [sidr] BGPSec scaling (was RE: beacons and bgpsec)

[Christopher Morrow <[e-mail]>]

On Fri, Sep 9, 2011 at 12:19 PM, Randy Bush <[e-mail]> wrote:
>>>     as a vendor friend says, if ipv6 deploys, insha allah, we're gonna
>>>     be upgrading those routers to do real v6 forwarding.  if it does not
>>>     deploy, you will be deploying massively bigger boxes to nat your ass
>>>     into hell.
>>
>> There are two possible results, it seems to me:
>>
>> 1. The cost of deploying IPv6 will "bury" the cost of doing BGPsec, so
>> that BGPsec essentially becomes "free" in the IPv6 upgrade.
>>
>> 2. The cost of deploying BGPsec will be significant enough that it can't
>> be "buried," in any other costs.
>>
>> The question is --which is true?
>
> as i have no data, any guess i make would be bullshit, would it not?

maybe what Wes is asking here is really:
"Could someone model the load on a router doing bgpsec, in a world of
bgpsec speaking devices?"

Something like, for a core network edge device (say sprint, C&W, TWTC,
UU/vzb,ATT an edge connecting device in their worst metro):
  o number of updates today/second (steady state and 'worst case')
  o projected growth of update stream (given historical data)
  o projected 'cost' (cpu cycles) of un-assisted bgpsec
  o projected RIB RAM size (use historical data to project forward)
  o projected beacons/second (which really just look like updates in
the update stream)
  o routing table size (projected forward from historical data)

It seems most of that data exists in one form or another, it seems
that running the math isn't "hard". There's a question of the validity
of the model... but that's always the case.

Wes, is this sort of thing what you're asking for?

-Chris