Re: [sidr] WGLC: draft-ietf-sidr-bgpsec-reqs

[Christopher Morrow <[e-mail]>]

On Fri, Nov 11, 2011 at 8:49 AM, Danny McPherson <[e-mail]> wrote:
>
> On Nov 11, 2011, at 8:19 AM, Christopher Morrow wrote:
>
> There's actually some research on this, I recall the number 'globally'
> as 1.2 avg packing... but internally, that may be different, of
> course.
>
> I'd be interested in a pointer to that Chris, if you could pass it along.

I had thought randy referenced the thing I am remembering, I'll ask
once I get to the meeting venue.
I think he's also pointed to some study work on the effect of
ingesting 1x prefix vs packed prefixes...

> The only quantitative analysis I've seen of this is here:
> <http://www.tcb.net/stuff/danny-ucla-pack.pdf>
> It's 3 months of data from 6 monitors.  The basic observation is that
> around 30% to 40% updates are packed, and these packed updates
> carry up to 80% of prefixes -- a density that seems to be fairly consistent
> across both iBGP and eBGP monitors.

neat.

> -danny