On Tue, Dec 6, 2011 at 10:56, Tao Sun <hisun...@gmail.com> wrote: > These are (at least the example 1) doable using RFC4191 and RAs. For the >> latter case how the default router for "default traffic" gets selected >> (independent of DHCPv6 or RAs) is a bit of an issue if all interfaces >> (APNs) are equal. >> > [Tao] The interfaces (APNs) are generally not equal, i.e., APN_IMS, > APN_Internet. >
Yes, and each one already has its own RA. Compared to AAA based on what grounds? >> > [Tao] Using radius with AAA is only used for few enterprise users in some > operator's network. > So what entity in the network decides what prefix to assign to a given user when he sets up, say, the Internet PDN? In the above (cellular oriented) example, I am not so sure. In cellular > (3GPP) networks the RAs are already used in per subscriber manner. > > [Tao] You need to update radius protocal and AAA specification in 3GPP to > convey the configuration information. The mandatory RA for prefix > assignment is supported not long time ago by network elements. Ohter > optional extension of RA such as RIO indeeds new requirement for network > elements and handset. > The RADIUS attribute for it already exists, it's called Framed-IPv6-Route: http://tools.ietf.org/html/rfc3162#section-2.5 . The handsets can easily support it if they are based on operating systems that support it. I know Linux does, so Android could use it. Don't know about OS X / iPhone, but on the other hand the iPhone doesn't support IPv6 on the mobile interface yet.
_______________________________________________ mif mailing list mif@ietf.org https://www.ietf.org/mailman/listinfo/mif
