Hi Uma

Thanks for this, in line below.

From: Uma Chunduri <[email protected]>
Sent: Friday, July 23, 2021 12:04 PM
To: David Allan I <[email protected]>
Cc: [email protected]
Subject: Re: [DMM] Additional questions/comments on 
draft-ietf-dmm-srv6-mobile-uplane-13

Hi Dave,

Very good list and I am sure there could be a bit more functionalities GTP 
overlay  has been supporting over a period of time *and* still currently being 
used in various parts of the cellular networks.

Replacing overlays may be a good goal if we can bring the same technical 
benefits different underlays (L2, L2.5, IPv4, IPv6) bring to the table in 
various parts of the networks, in this case cellular deployments. We(IETF)  
should develop tools towards the objective but I suspect we are there to 
replace any overlay (if this is the central goal) with a single underlay 
mechanism. Dino presented multiple times on this topic in various 
WGs/pre-bofs/BoFs and also asked to "get over with it" (a thin overlay).

One of the things that we all should ask ourselves is - what overlay 
functionalities we want to bring to underlay (L3/L2.5) as opposed to complete 
replacement of the overlay. IMO, this is a better question than "replacing".

Appreciate your thoughts and views on this.

DA> I think we are in a situation where what we need to bring forward in a 
solution mirrors what is there ESPECIALLY if the goal is a transparent stunt 
double to the higher level 5G System.  From what I can tell 3GPP already has a 
bit of a winnowing process if not from release to release, at least from 
generation to generation.  So we are not in a situation like for example, the 
early days of pseudowires where the deployed protocol landscape was massive and 
triage was definitely called for.   What is more then there was complete 
divergence in systems as you moved up the stack whereas transport for 5G serves 
a common higher level system.

The sorts of features we are seeing in the larger 3GPP picture AFAIK are more 
recent additions for URLLC, CIoT etc.  So picking and choosing based on 
convenience is not an option; the problem space is not a matter of choosing 
what to support, but in replicating the functionality exists and having similar 
extensibility.

A further note is that GTP is not in isolation, and some functionalities are UE 
to UPF (e.g. QMP),  so interworking between “whatever” and RLC etc on the radio 
link is also table stakes.

At least that is the view from here
Regards
Dave



On Fri, Jul 23, 2021 at 10:50 AM David Allan I 
<[email protected]<mailto:[email protected]>>
 wrote:
Hi:

Looking over this draft it strikes me that IF you are discussing a GTP 
replacement, there are a few unaddressed gaps...or at least I see no mention of 
them.

-  adding X2 to the list of interfaces to be supported
-  support for end-marker for handoff stream synchronization (probably could 
fit the unused bit in Args.Mob.Session but frankly I have not studied this in 
detail)
-  support for QMP (quality measurement protocol), which I believe needs to 
also extend across the radio interface so is not simply local to the GTP domain 
and can be "stunt doubled"
-  support for paging policy presence/indication

If this has come up elsewhere, let me know

Cheers
Dave


_______________________________________________
dmm mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/dmm
_______________________________________________
dmm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmm

Reply via email to