Re: [Pana] What to do with I-D: draft-ietf-pana-ipsec

["Joseph Salowey (jsalowey)" <jsalowey at cisco.com>]

If the use of the key is entirely within the PANA authenticator and PANA
specifications then using the MSK is OK.  If the same key is going to be
used in specifications independent of PANA (802.11, etc) or used
somewhere other than the authenticator then the MSK may not be a good
choice.  

> -----Original Message-----
> From: pana-bounces at ietf.org [mailto:pana-bounces at ietf.org] On 
> Behalf Of Jari Arkko
> Sent: Wednesday, April 15, 2009 7:06 AM
> To: Alper Yegin
> Cc: yohba at tari.toshiba.com; pana at ietf.org; Basavaraj.Patil at nokia.com
> Subject: Re: [Pana] What to do with I-D: draft-ietf-pana-ipsec
> 
> I think I agree with what Alper is saying below. It is 
> obviously important to have separated keys for PANA itself 
> and the per-packet protection (such as for IPsec). But given 
> the definition of the keys used for PANA in the RFC, I think 
> it is possible to have other MSK-derived keys for IPsec.
> 
> Jari
> 
> Alper Yegin wrote:
> >> As
> >> for draft-ohba-pana-pemk-02, it specifies (as does 5191) 
> the use of 
> >> the MSK which is a _really_ bad idea IMHO -- the EMSK 
> should really 
> >> be used instead.
> >>     
> >
> > Why so?
> >
> > Secure association protocols have been using MSK-driven keys.
> > And it makes sense, as MSK is what NAS knows (not EMSK).
> > I have no idea what value using EMSK has, but the obvious 
> cost is to 
> > impact the AAA deployment between the NAS and AAA servers. 
> Today AAA 
> > protocols deliver MSK, not EMSK or any of its children.
> >
> > Alper
> >
> >
> >
> >
> >
> >
> >   
> 
> _______________________________________________
> Pana mailing list
> Pana at ietf.org
> https://www.ietf.org/mailman/listinfo/pana
>