Re: [v4tov6transition] V0.4 Agenda for WebEx meeting on Sep 20/21

[jouni korhonen <[e-mail]>]

Hi,

I had a quick look at some of these documents. See comments below.

On Sep 20, 2010, at 6:18 AM, Tina TSOU wrote:

[snip]

>  
> 7. Use Case for Mobile scenarios
> draft-zhou-v4v6tran-mobile-use-case
> Presenter: Cathy Zhou (one or two slides and 3 minutes to explain it)
> Questions for clarification (5 minutes) (More slides can help when the authors answer the questions in these 5 minutes.)

The use of GTP-U vs PMIPv6 terminology gives on impression that those are exchangeable just like that. However, GTP also has the control part GTP-C. Therefore, the text should be more careful when it means GTP as a whole (both -U & -C), or just  -U or -C part of it.

The draft could actually reference to draft-korhonen-v6ops-3gpp-eps on parts where it explains the basic EPC/GPRS architecture. That part is rather well covered in the mentioned draft.

>  
> 8. IPv6 Transition Guide For A Large Mobile Operator
> draft-tsou-v4v6tran-mobile-transition
> Presenter: Tom Taylor (one or two slides and 3 minutes to explain it)
> Questions for clarification (5 minutes) (More slides can help when the authors answer the questions in these 5 minutes.)
> 

o Section 1:

   strategies and tools for migration. 3GPP has decided to use Gateway-
   Initiated DS-lite [ID_GI_DS_lite] as the primary tool for subscriber
   migration.  Details as far as they have been thought out are provided 
   in [3GPP_TR_23_975].  However, they cover only a small part of the
   total problems.

I think this statement is making a forward step from 3GPP process point of view. GI-DS-Lite is still in a public Technical Report.. there is no any mandatory specification work done in 3GPP yet.

o Section 2.2:

      A key use of this DNS system is to allow the Serving Gateway to
      locate a PDN Gateway providing access to the core network that the
      subscriber will use.  This may be operator's own core network or
      the subscriber's home network.

I assume the above refers to 3GPP EPC. There the Serving GW does not do any DNS queries to locate the PDN-GW. It is MME that does it and locates a PDN-GW for a Serving GW.

And a PDN-GW does not provide access to core network for the subscriber. In 3GPP terminology that is more like an access to a PDN.

o Section 2.4:

   3GPP specifications for mobile devices have required dual stack
   support for at least a year (i.e., as of Release 8.)  The operator

As far as I know, "required" is too strong. Dual-Stack in devices is still an optional feature in Release-8 and it is possible to ship single IP version capable stacks e.g. IPv6 only device. If the device stack does both IPv4 & IPv6 then IPv4v6 bearer is required.

   user devices and sharing public IPv4 addresses between multiple users
   using some system of block port allocation as proposed in
   [ID_natx4-log-reduction].

What about draft-ietf-dime-nat-control ?

   DNS access is an issue with Gateway-Initiated DS-lite.  The original
   DS-lite proposal had a point (the B4) at which all DNS queries could
   be intercepted and sent to an IPv6 DNS.  From here IPv4 queries could
   be forwarded to an IPv4 DNS, or the IPv6 DNS could maintain both AAAA
   and A records.  It is not so obvious that such interception can be

In DS-Lite the DNS issue was different compared to this i.e. a provisioning and IP transport version issue. I don't see the relation to GI-DS-Lite here. In GPRS/EPS the mobile can be provisioned with both IPv4 and IPv6 DNS server addresses during bearer setup, and these DNS servers are then both behind (S)Gi. Are you mixing IP transport version and DNS configuration information here..?


- Jouni

[snap]