Hi,
some comments inline.
3gv6-bounces <at> ietf.org 写于 2010-04-01 下午 11:26:23:
> Hi, Frank,
>
> If the prefixes assigned to one PDN connection are aggregable, it
is
> fine. The problem is we have to carefully deal with the possible
> gaps that may be left after several times PDs.
> I think it is better if we can assign one shorter prefix to UE in
one time.
XY: Yes, agree.
>
> I don't think R11 needs more complex alternatives. If we can figure
> a good solution for R10, R11 absolutely can use this solution if the
> requirement is not changed.
>
> I will check your contributions to 3GPP and give you feedback in a
few day.
>
> Using different APN for different prefix length is not practical.
> Usually, We do not set too many APNs in the field network. It is not
> convenient for the UE to change one APN to another if it wants a
> different length prefix. Usually, differnt APN is set for different
> service. UEs connected to the same APN may want to get different
> length prefixes due to differnt networks behind them.
XY: Yes. It is not practical that
using different APN to represente different length prefix service. But
defining a particular APN can be used to help UE to select the PDN which
providing shorter prefix service or supporting DHCPv6 PD. Anyway, not every
UE needs shorter prefix, neither of the P-GW.
>
>
Xingyue
ZTE Corporation
> --------------------------------------------------------------------------------
>
> Zhenqiang Li
> 13911635816
> Department of Network Technology
> China Mobile Research Institute
> 2010-04-01
>
> --------------------------------------------------------------------------------
>
> 发件人： Frank Brockners (fbrockne)
> 发送时间： 2010-03-30 15:50:07
> 收件人： lizhenqiang <at> chinamobile.com; suresh.krishnan <at> ericsson.com
> 抄送： youni.korhonen <at> nsn.com; dhcwg <at> ietf.org; 3gv6 <at> ietf.org;
> Dhananjay Patki (dhpatki)
> 主题： RE: [3gv6] about Prefix delegation in 3GPP EPC networks
>
> Hi Zhenqiang Li,
>
> please see some thoughts inline.. (Suresh will probably add more)
>
> > -----Original Message-----
> > From: 3gv6-bounces <at> ietf.org [mailto:3gv6-bounces <at> ietf.org] On
Behalf
> Of
> > lizhenqiang <at> chinamobile.com
> > Subject: [3gv6] about Prefix delegation in 3GPP EPC networks
> >
> > Dear Suresh,
> >
> > Nice meeting you at IETF 77.
> >
> > First of all, I agree with you that it is good for a UE to use
a
> > shorter prefix to meet the situation, rather than using multiple
/64
> > prefixes.
>
> ...FB: there's been several discussions at this past IETF on this
> particular topic - and the approach which seems to have wide support
is
> that for Release 10, one would support a model where one PDN connection
> would continue to be represented by a *single* prefix. This prefix
could
> be shorter that /64. This means that we can still hand out multiple
> prefixes to the UE, but it would always be possible to aggregate all
of
> them into a single prefix. We'd continue using SLAAC for assigning
the
> /64 for the UE (or better PDN-Connection-link) and would hand out
the
> remainder of the "shorter prefix" using DHCP PD. This will
keep changes
> to the current architecture to a minimum (incl. PCC will need to deal
> with only a single prefix in its rules). More complex alternatives
would
> be explored in future releases (i.e. Release 11 and onwards).
>
> I'm working on updating the contributions S2-101176, S2-101177,
> S2-101178 that were submitted "for information" to the last
3GPP SA2
> meeting to reflect these changes (which are only small deltas to the
> current versions, so hope to have them available for discussion in
the
> next few days) (You can find the submitted version at
> http://www.3gpp.org/ftp/tsg_sa/WG2_Arch/TSGS2_78_San_Francisco/Docs/)
>
> >
> > Some additional comments about your draft:
> > 1. Where to store the prefix length that you reserved for each
UE?
> > 2. How to validate the request from UE to avoid address abusing?
>
> ...FB: The expected approach is that shorter prefixes would be handled
> as a capability of the APN. This way you also won't have any issues
with
> "abuse" - each UE (for a particular APN supporting shorter
prefixes)
> could use DHCPv6 PD to request portions of the prefix.
>
>
> > 3. The proposed method in your draft seems address wasting, because
> you
> > reserve /x (x < 64) prefix for each UE. It is better
to store the
> > minimum length that a UE can apply, and assign the proper lengh
to UE
> > based on the available adrress pool, the minimum length the UE
can
> > apply, and the length the UE actually requests.
>
> ...FB: If we do things on a per APN basis (as mentioned above), one
> would expect that only those UEs who need shorter prefixes would use
the
> particular APN, so the amount of "wasted" addresses would
be relatively
> small. That said, the approach would just be for Release 10 (we just
> have two more meetings to get this off the ground) - more optimal
> approaches could be studied in future releases.
>
> Regards, Frank
>
> > 4. Are there any other alternatives besides DHCPv6-PD? I listed
some
> > potential solutions in a 3GPP document. What is your opinion
to them?
> >
> > Best Regards,
> > Zhenqiang Li
> > 13911635816
> > Department of Network Technology
> > China Mobile Research Institute
> > 2010-03-26
> > _______________________________________________
> > 3gv6 mailing list
> > 3gv6 <at> ietf.org
> > https://www.ietf.org/mailman/listinfo/3gv6
> _______________________________________________
> 3gv6 mailing list
> 3gv6 <at> ietf.org
> https://www.ietf.org/mailman/listinfo/3gv6
RSS Feed