[e2e] It's all my fault

[Lachlan Andrew]

On 15/05/07, Jon Crowcroft <Jon.Crowcroft at cl.cam.ac.uk> wrote:
> if we took ALL routing out of routers, life _would_ be simpler -
>
> you'd only have yourself to blame as an end user
> for shooting your self in the head when you can't reach somewere
>
> running path computation on boxed _designed_ to do computation
> and forwarding on boxeds designed to switch packets fast
> just sounds like a perfectly reasonable idea to me

I'm in favour of loose source routing, but don't agree that routers
shouldn't route.

The simplicity and distributed computation of having a "default next
hop" would be lost, as would the ability to route around failures.

Loose source routing gives the best of both worlds:  users get to
explore multiple paths and choose the best (as done explicitly by
Akamai and implicity by BitTorrent), but don't have to specify the
paths in more detail than they know.

Regarding "denial of service", if we regard sending packets to where
users want them to go as a "service", disabling source routing is
itself a massive denial of service :)   Fortunately, if source routing
is in the standards and implemented by enough systems, then users can
route around those denials of service, but if it is not even in the
standards, then all seems lost.

Perhaps a compromise would be to reduce the number of intermediate
hops that can be specified from 40 to say 2.  That reduces the
"traffic multiplier" available for DoS, but allows users to select
between a handful of paths.  Two or three paths is enough diversity to
get a "pretty good" route if the default BGP route is temporarily
congested.

Cheers,
Lachlan

-- 
Lachlan Andrew  Dept of Computer Science, Caltech
1200 E California Blvd, Mail Code 256-80, Pasadena CA 91125, USA
Phone: +1 (626) 395-8820    Fax: +1 (626) 568-3603