Re: [v6ops] new draft: draft-taylor-v6ops-fragdrop

[Brian E Carpenter <[e-mail]>]

On 18/10/2012 08:07, joel jaeggli wrote:
...
> The focus on core devices is misguided imho... The services edge of my
> network where both connections terminate and policy is applied may be
> substantially higher capacity than the backbone of many networks it is
> exposed to this issue. Given that I am an end-site all traffic carried
> on a backbone is either coming from or in bound to an end system on our
> network.

Yes, and as far as I can tell the problems arise with site-boundary
devices that attempt to indulge in deep packet inspection and are
stymied by extension headers and/or fragments. Load balancers have
similar difficulties. These devices do violate the letter or spirit
of RFC 2460, but that's life.

(The fact that RFC 2460 does not spell out behaviour for extension headers
not mentioned in RFC 2460 is why I wrote draft-carpenter-6man-ext-transmit.
But that will not magically change the code in firewalls.)

Core routers just route packets. ECMP/LAG might not work optimally in some
cases, but that doesn't break e2e connectivity.

    Brian