RE: 3gpp-analysis-05: miscellaneous non-critical issues

[Pekka Savola <pekkas@netcore.fi>]

On Sun, 28 Sep 2003, Soliman Hesham wrote:
>  > Do you have an alternate wording suggestion in mind which 
>  > would keep the
>  > essence of the original text (including "don't tunnel from 
>  > UE")?  This
>  > doesn't and is unacceptable to me..
> 
> => I didn't ask to have "don't tunnel from the UE"
> in the text. I was commenting on the new text.
> I don't think the draft should include opinions on how 
> easy it is to upgrade to V6 and whether that involves
> SW or HW upgrades. It's not useful. 

It seems very useful to me, although the wording may not be optimal.  If 
upgrading to v6 is relatively easy, we can recommend upgrading to IPv6.

>  > These are two quotes are intentional.  IPv6 support will
>  > work when the
>  > visited networks also have upgraded to this minimal IPv6 
>  > support.  
> 
> => (drifting from the original comment but I'll answer 
> this as a side issue). Supporting PDP type V6 is not a 
> minimal support because it implies that the network behind
> the SGSN supports v6, i.e. GGSN, DNS ...etc.

When your network infrastructure includes hundreds of boxes, enabling v6 
support in about 3 of them could be seen as very minimal.

Note: I think it might be useful to try to describe in the document what 
constitutes a bare miminal IPv6 support.. just spell it out.  DNS 
certainly is *not* required.

>    That's 
>  > also expected to happen during the early phases of 
>  > transition.  If the 
>  > operator doesn't care about IPv6 users at all, the users can 
>  > just stay out 
>  > of those networks.
> 
> => The point is: Your operator cares about v6 and wants 
> to offer you the services that IPv6 enables 

Right so far..

> while you're
> roaming. 

If your SP wants to enable services while you're roaming, your SP needs to 
convince those networks it has roaming agreements with to enable v6? 

> It is not relevant how the visited operator feels
> about v6. 

I fear this is very relevant.  Your home operator can't really offer what 
it can't provide.  So, it should not *guarantee* you IPv6 services while 
you roam, as simple as that..

> So to work around this, you need to use something
> to enable you to tunnel to a router in the home network. 
> IMO ISATAP fits nicely in this model. 

That's a detail at this point, but this is certainly at odds with the 
ISATAP security model.

> But the point is, 
> requiring an operator to support v6 will not necessarily
> eliminate tunneling when the UE roams to another network
> (whose managers don't want to support v6 and will implement
> v6ops recommendations). 

So, don't use v6 in those networks then?

>  > We discussed previously several other options, such as 
>  > trying to open the
>  > IPv6 PDP context to an operator (if there are many) which in 
>  > fact supports
>  > IPv6.  This would be useful in the case that 3G + IPv6 gains market
>  > momentum as there are typically multiple networks in a 
>  > region one could
>  > roam to.  I'm not sure how many of those ended up in the document.
>  > 
>  > The bottom line is that as we deploy dual stack UE's, when IPv6 isn't
>  > supported in the visited network, one can just use IPv4 if 
>  > all the tricks
>  > fail.  
> 
> => So , "other tricks" do not include tunneling? Why not?

That'd de-facto require implementation of a number of transition
mechanisms in UE's, with caveats those entail, making it much more
complex.
 
>   No tunneling needed at the UE.  Further, tunneling at 
>  > the UE is 
>  > counter-productive, as it doesn't give the *visited networks* _any_ 
>  > incentive to deploy IPv6 in the first price, which is the number one 
>  > priority.
> 
> => But in this case the visited network doesn't care about 
> v6. I'm not sure how eliminating the use of V6 through 
> a tunnel will give the visited network an incentive to
> deploy v6.

For example, if the customers select a different operator when they roam
because the one they tried first doesn't support v6 ("v4-only operator
loses customers and thus money"), I guess the operator which didn't
support v6 would start to see some benefit in deploying v6 if the number
of users warrants that.  There could also be some PR gains/losses involved
here.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings