Le 03/05/2018 à 15:55, Sri Gundavelli (sgundave) a écrit :
It is probably the only reason at this time that makes Mobile IP still
necessary.
Not really. You will have the same issue with Mobile IP.
Static allocation implies the UE’s session is anchored on a gateway node
which is the topological anchor for that address block.
Well, one can have one own's HA (not cellular network's) to manage the
static prefix allocated to the UE, and the cellular network to assign a
variable prefix in RA.
Unless, the assigned prefixes are dynamically programmed (ignoring other
issues), UE’s session always needs to anchored on the same node.
Yes, the HA.
Alex
Sri
On 5/3/18, 12:06 AM, "Alexandre Petrescu" <[email protected]>
wrote:
Le 02/05/2018 à 16:29, Sri Gundavelli (sgundave) a écrit :
I can agree that the possibility with RADIUS/DIAMETER permits to
alocate
a stable prefix in RA to a UE. However, I have never seen it in practice
in a cellular network.
Enabling static IP allocation by default has a scaling issue. The IPv6
prefix that is allocated to the UE is part of an aggregate block
configured on a given PGW node. Reserving IPv6 prefixes from that block
makes that one PGW node as the anchor for that UE. Operators loose
flexibility with respect to gateway assignments.
Yes, I agree.
If you look at some of the work that went in 3GPP Rel 10 timeframe, it
was
about enhancements to gateway selection logic based on number of
access-network parameters. It allowed operators to allow gateway
selection
based on proximity, policy and other parameters. Now, any time there is
an
IPv6 prefix reservation that flexibility is lost, as that creates a
Gateway/Subscriber stickiness, and potentially resulting in uneven
distribution of subscriber session across all the gateways. So, this is
an
operational problems and may be v6ops is the right group for such
discussions.
I can agree - it is an operational issue.
It is probably the only reason at this time that makes Mobile IP still
necessary.
Alex
Sri
On 5/2/18, 12:28 AM, "Alexandre Petrescu" <[email protected]>
wrote:
Le 25/04/2018 à 05:00, Sri Gundavelli (sgundave) a écrit :
Hi Alex,
I cannot comment on the supported network/service configuration in any
operator's network. But, I’d think the allocation of stable /64’s is
similar to static IPv4 (/32) address allocations that are supported in
many operator networks today. There are also RADIUS / DIAMETER
attributes
such as Framed-IPv6-Prefix ..etc which can be used for obtaining
statically configured values by PGW. So, IMO, its very much possible
to
allocate static IPv6 Per-UE prefixes for the UE. Its also possible to
allocate static IPv6 prefixes for the networks behind UE. IMO, this is
just a configuration and operators can surely do this today.
But, I am not sure where we are going with this?
I am asking because I would like it to happen as you describe as being
possible.
I can agree that the possibility with RADIUS/DIAMETER permits to
alocate
a stable prefix in RA to a UE.
However, I have never seen it in practice in a cellular network.
When I connect I always get a different IPv6 prefix in RA.
Where are we going with this? Let us go towards identifying first if
_anybody_ (any cellular network) allocates a stable IPv6 prefix in RA
to
UE. It may be someone does allocate such, as it may be nobody does.
In the first case, then I will suggest my operator to do likewise.
In the latter case, then let us go in a direction where we understand
_why_ operators dont implement stable IPv6 prefixes per UE.
Alex
Sri
On 4/24/18, 7:31 AM, "Alexandre Petrescu"
<[email protected]>
wrote:
Hi Sri,
Is there an operator today that allocates a stable /64 in RA to User
Equipment? (resists re-connection)
Alex
Le 26/03/2018 à 07:14, Sri Gundavelli (sgundave) a écrit :
Alex:
This is a good point. Yes, there is DHCPv6 prefix delegation support
in
3GPP architecture for supporting mobile router use-cases. This is
essentially for delegating prefixes for the networks attached to the
UE.
This was introduced in Rel-10 by cisco. I have not followed the
recent
SA2
discussions and I do not know if MR support based on DHCPv6 will
continue
to be supported or not, and if they have considered the alternative
options for supporting the same. I think we can certainly ask that
question, but I also wonder if the coloring is specific to the PDU
session, or if its broadly applicable for all UE address/prefix
assignments.
Sri
On 3/23/18, 2:48 AM, "dmm on behalf of Alexandre Petrescu"
<[email protected] on behalf of [email protected]>
wrote:
Le 22/03/2018 à 18:49, Liaison Statement Management Tool a écrit :
[...]
SA2 would like to point out that among the four mechanisms for
address configuration delivery mentioned in your LS reply (i.e.
DHCPv4, DHCPv6, IPv6 ND and IKEv2) only the IPv6 ND mechanisms,
and
in particular the Router Advertisement message, seem to be
applicable
in the 5G System architecture in the specific context of
Multi-homed
IPv6 PDU Sessions.
Please tell SA2 that current 4G cellular networks are specified to,
and
do use to some extent, DHCPv6 Prefix Delegation to assign a prefix
to
an
end node like an IoT Router.
A /56 prefix delivered to the end node should have the same
capabilities
as an address. We also want that /56 prefix to be more stable, or
less
stable, etc.
I dont understand why 5G System architecture excludes DHCPv6 from
the
list of applicable address configuration delivery.
I understand why 5G System architectures prefers ND - it is for
addresses.
Alex
With respect to the following question in the IETF¹s reply LS:
We also like to point out that, all though
the LS
statement explicitly refers to both IPv4 and IPv6 address types,
however
it only mentions about (RA) (IPv6
ND implied) as the mechanism for address
property
delivery. It is to be noted that the approach of delivering
coloring
meta-data in RA messages will most
likely be to limited to IPv6 address/prefix
types
and
will not be extended to IPv4 addresses. If this capability is
required
for IPv4, we may have to possibly
extend DHCP protocol(s).
We request 3GPP to clarify if the Ask is
explicitly
for IPv6, or if its for both IPv4 and IPv6 address/prefix types.
SA2 would like to clarify that the request is explicitly for IPv6.
SA2
discussed the example documents that were referenced in your LS
reply
and concluded that the following draft seems to be the most
promising
candidate for the problem under discussion in this correspondence:
https://www.ietf.org/id/draft-feng-dmm-ra-prefixtype-01.txt.
SA2 would like to kindly ask IETF DMM working group to keep SA2
updated
of the work on the subject of including property meta-data in IPv6
ND
address assignment procedures for potential use in the 5G System
to
indicate the mobility property of additional IPv6 prefixes
assigned
as
part of the Multi-homed IPv6 PDU Session functionality.
2 Actions
To IETF DMM working group:
ACTION: SA2 would like to kindly ask IETF DMM working
group
to
keep SA2 updated of the work on the subject of including property
meta-data in IPv6 ND address assignment procedures for potential
use
in
the 5G System to indicate the mobility property of additional IPv6
prefixes assigned as part of the Multi-homed IPv6 PDU Session
functionality.
3 Dates of next TSG SA WG2 meetings
TSG SA WG2 Meeting 127 16 - 20 Apr 2018 Sanya, CN
TSG SA WG2 Meeting 127-Bis 28 May 1 Jun 2018 Newport Beach, US
Attachments:
S2-182967_was2844_LS_IETF_SSC3
https://www.ietf.org/lib/dt/documents/LIAISON/liaison-2018-03-22-3g
pp
-t
sg
sa-sa2-int-6man-dmm-ls-on-indicating-service-continuity-usage-of-th
e-
ad
di
tional-ipv6-prefix-in-router-advertisement-attachment-1.docx
_______________________________________________
dmm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmm
_______________________________________________
dmm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmm
_______________________________________________
dmm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmm
