I have two general concerns about the use of IPv6 in 3GPP
(i.e. things not yet in the document).
First, as it has already been discussed, there can be two kinds of UE
(the customer equipment):
- 3GPP view: some kind of smart handset or PDA which has the 3GPP
stack with radio, ... This kind of devices has *not* enough
human interfaces to be able to do something interesting IMHO but
this is another debate (:-).
- our view: a laptop connected to a handset or a dedicated board.
This laptop has a regular IPv6 stack so the attachement device
has to present the 3GPP connection as a standard network (i.e.
Ethernet (what is done by IEEE 802.11 for instance) or PPP with
IPv6CP).
I believe everybody agree about what not to do in the second case...
The second concern is a bit different, GGSNs will be the point of
connection to an IP world (the Internet, an Intranet) only if the
regulation bodies are not involved. We already know the vertical
integration in a walled garden dream of telecom operators but
fortunately regulation bodies (like ART in France) are here in
order to require a real open market for IP services. This is not
new and this already happened for ADSL, so I believe we'll get
similar solutions.
If the point of connection to IP is a NAS (Network Access Server)
of an ISP, the PDP type of choice should be PPP, just because
PPP/Radius provides a good network access control. Note that
this solves:
- the UE issue because PPP is a very well known protocol
- the network access control issue because PPP has good
authentication-authorization-accounting features (PS:
IP network access control, the radio network access control
should be the job of the SIM based stuff. They have to be
different because the operator and the ISP are not the same entity)
- the GGSN routing issue because if addresses are allocated by ISPs,
the GGSN has to manage a zillion of host (or /64) routes.
- the GGSN to ISP attachement (Gi) because there is no reason to
not reuse the current ADSL solution (ATM (sic!) to BASs).
etc...
Two small (other) remarks:
- in order to justify the allocation of /64 in any cases,
we should require the support by NASs of UE temporary addresses
(so the recommendation 3 will become a strong one).
- the dialup document (draft-itojun-ipv6-dialup-requirement-02.txt)
*applies* so:
* /128 MUST NOT be considered (/64 is the minimum)
* allocation model discussion is pertinent so the UE SHOULD be
by default considered as to be a router
Regards
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
