Re: [GROW] WGLC: draft-ietf-grow-simple-leak-attack-bgpsec-no-help

[Christopher Morrow <[e-mail]>]

On Wed, May 14, 2014 at 3:59 PM, George, Wes <[e-mail]> wrote:
> 1) route leaks are [definition, see text suggestion in previous message]
> 2) there are ways for this sort of leak to be used as a MITM attack
> [example from your draft]

I would actually argue that the general case of 'route leak' causes
things which are potentially more (broadly) harmful:
  latency increases
  loss increases
  blackholing

Sure, MiTM is a side effect of this, but really as soon as you cross
AS boundaries (even one you WANT to cross) you are open to MiTM
attacks. For GROW I would think the relevant bits are the latency,
loss, blackholing concerns.

MiTM gets good press, but isn't really required to discuss the actual
problem of 'oh crap why did my packets take a left turn THERE??'

> 3) these leaks occur very frequently [data citations from your draft]
>         3a) but not all are malicious, some are misconfigurations, some are
> intentional
> 4) routing hygiene helps to prevent, but not eliminate [discuss gaps]
> 5) BGPSec doesn’t fix, because it can only secure BGP attributes, and BGP
> has no semantics to convey intent or this type of inter-AS propagation
> boundary policy

(I agree with the rest of wes' points... I'm not sure it benefits
anyone in GROW land to hammer on bgpsec, especially when things like
rpki deployment would help stop leaks, if used in conjunction with
route filtering practices)

-chris
'just a guy' hat on.