Re: [saag] RADIUS is deprecating MD5

[Peter Gutmann <[e-mail]>]

Alan DeKok <[e-mail]> writes:

>Um... that is very much not what RADIUS is.

I'm not talking about RADIUS as RADIUS, I'm talking about RADIUS as a
substrate for the next half-dozen layers up.  In other words RADIUS is present
purely because the spec says there needs to be something between UDP below and
EAP-over-RADIUS above, which in turn is present purely because the spec says
you need that before you get to one of several types of TLS tunnel carrying
the actual payload you care about, inside the tunnel.

So it doesn't matter if, as used in that situation, RADIUS uses a Fletcher
checksum for "security" or prepends the password to the first packet sent
because it's acting only as a substrate for all the other layers, including
the ones providing the security.  That's the case I mentioned, user =
"anonymous", password = whatever (I've seen things like "anonymous",
"password", "dummy", and others) because RADIUS isn't doing anything except
meeting a requirement that it be present.

>I wouldn't really say there's "nothing worth attacking".

Probably best for another thread, but if there aren't any major attacks
against it occurring, as you mentioned previously, then it seems like
attackers don't consider it worth attacking.  That was the case with DKIM
signing, it wasn't defending against anything that attackers cared about so
none of them bothered exploiting the very weak keys used.

Peter.