[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: R41 in draft-ietf-v6ops-cpe-simple-security-07



I did not say anything about layer 2 or 3 or UPnP. I'm not interested in holding a beauty contest for the various ways to do firewall control. The point I'm
trying to make applies to all of them.

That being said, we seem to be in agreement for having authentication as an
option for any firewall control.  I would add that to R41.

Mark

On Aug 5, 2009, at 11:04 AM, Rémi Denis-Courmont wrote:

Le mercredi 5 août 2009 20:01:33 Mark Baugher, vous avez écrit :
I expect to see more providers and vendors
ship their wi-fi products with privacy turn on by default, and who
therefore expect their customers to authenticate to a wi-fi/router as
part of the process.  So I think your statement above is shortsighted
for at least this reason.

If the host operating system is authenticated to the CPE at layer-2, why the
heck should it re-authenticate at layer-3? To me the big problem
_specifically_ with UPnP is that any random application can use it and
complete non-sense stuff with it like redirect any port to any IP and port.

As a counter-example, if I recall correctly, ALD can only redirect ports to the host that request it, and requires system privileges on the host (as it
runson ICMPv6).

There is innovation taking place in home network access controls - WPS
is one attempt by vendors to provide usable security in an unmanaged
environment.  I know of a few other organizations that are designing
security into their home networking services and not interpreting
"unmanaged" as meaning "insecure". It would be shortsighted to design a firewall control protocol that does not at least have access control
as an option.

I already said I am all for having authentication as an option. But
reallistically, I doubt it can nor should be enabled by *default* on
*unmanaged* networks.


--
Rémi Denis-Courmont
http://www.remlab.net/