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

[Jakob Heitz <[e-mail]>]

10u to validate an incoming prefix.
How long to validate an incoming path?
How long to generate/sign an outgoing path?
To one neighbor? To all neighbors?

--
Jakob Heitz.


On Sep 12, 2011, at 7:33 AM, "Randy Bush" <[e-mail]> wrote:

>> Q2: Is it well rooted in the historical measurement data for its
>> projections into the future?
>> 
>> Q3: Are additional assumptions made for projecting into the future
>> reasonable? In this instance, are the assumptions regarding the BGPSEC
>> take rate reasonable?
> 
> as humans and as engineers, we are well known for poor predictions of
> the future.  we each have a model and they differ widely.  and if you
> extrapolate, especially with a complex model, i become very suspicious
> of any and all of your data.
> 
> this is why the data i give on rpki-rtr is what we have measured and
> only what we have measured, 10u on a gsr/xr to validate one incoming
> prefix, 4-7 seconds on gsr/xr, 7206/ios, and juniper m7 and m240 to load
> 360k roas built to map the current route views table.  this is from a
> freebsd 8.2 server over both local gige and wan with 63ms rtt, seems to
> make little difference.  we have not yet been able to make finer grained
> measurements,
> 
> you are welcome to extrapolate all you wish.  i do not wish to.  i have
> no idea what the timing would be with twice that many roas.  i would
> guess it would take more time, but i really have no idea, it could be a
> flat constant!
> 
> randy