Re: [Mipshop] transient binding draft update

["Ahmad Muhanna" <amuhanna at nortel.com>]

Hi Vijay,
Thanks for your detailed comments. Please see comments inline.

Regards,
Ahmad

> -----Original Message-----
> From: Vijay Devarapalli [mailto:vijay at wichorus.com] 
> Sent: Thursday, September 10, 2009 6:43 PM
> To: Muhanna, Ahmad (RICH1:2H10); Marco Liebsch; Mipshop
> Subject: Re: [Mipshop] transient binding draft update
> 
> Hi Ahmad,
> 
> On 9/4/09 1:17 AM, "Ahmad Muhanna" wrote:
> 
> >> The "static configuration" at the LMA actually starts accepting 
> >> uplink packets from the nMAG only after receiving a PBU from nMAG.
> > [Ahmad]
> > This actually may explain why you insist on the text above 
> under the 
> > previous paragraph. It is very interesting! First of all, is this a 
> > new usecase that you would like the transient BCE draft to 
> address?! 
> > IMO, it is a little late but we can consider that if we 
> understand the usecase.
> > 
> > Though, it will be helpful, if you take us through the 
> details of this 
> > usecase and what you are trying to optimize and achieve by 
> using the 
> > tBCE in here. Also, whether it is a new addition on top of RFC5213?
> 
> There is no new use case. The use case, IMO, has always been 
> that the LMA should accept uplink packets both from the pMAG 
> and the nMAG during a handover for a certain amount of time. 
> How you achieve this can be different.
> 
> In the transient binding cache mechanism, there is an 
> explicit signal from the nMAG to the LMA to accept uplink 
> packets from the nMAG even though the BCE points to the pMAG. 
> A subsequent PBU message from the nMAG updates the binding 
> cache entry to point to the nMAG. After that the LMA stops 
> accepting the uplink packets from the pMAG.
> 
> In the static configuration case, the PBU from the nMAG would 
> update the binding cache entry at the LMA immediately with 
> the proxy CoA being set to the nMAG. But, the LMA would be 
> configured to accept uplink packets from the pMAG for a short 
> duration (configured on the LMA). When the timer for this 
> duration expires, the LMA will accept uplink packets only 
> from the nMAG.
[Ahmad]
I see what you are saying here. But, usecase 1 is a little different. In
usecase 1, we are not worried about the uplink data traffic from the
pMAG after the 1st PBU with the transient option included. We are
worried about the downlink traffic. In other words, in usecase 1, if we
allow the LMA to update the BCE with nMAG pCoA immediately, the LMA will
start forward downlink traffic to the nMAG immediately and that is a
problem for usecase 1. Because, the link between the LMA all the way to
the base station in the downlink direction is NOT setup yet.

In usecase 1, IMO, 3GPP folks did an excellent job designing a very
reliable well thought handover mechanism.
1. When the MN moves to the target access network (nAN) with an active
data session, the MN does not need to worry about uplink traffic nor
downlink and it continues to send uplink traffic as normal. That is why
as soon as the MN moves to the target access network, the uplink
direction of the link from the base station all the way to the LMA is
ready. 

2. In order to avoid any packet loss in the downlink direction, all
downlink traffic continues to go to the pMAG to the pAN and from the pAN
to the nAN and then to the MN.

3. while the data traffic is moving as in 1 and 2, the network is
working on setting up the link for the downlink direction from the LMA
all the way to the nAN. 

4. When the link in the downlink direction is complete and ready, the
LMA switch the tunnel completely to point to the nMAG and nAN.

5. After the LMA switches the tunnel completely, there is no uplink
traffic in-flight to worry about from the pMAG because the uplink
traffic has been going through the nMAG directly to the LMA for
sometime. Only, in-flight downlink traffic which is sent to the pMAG.
But, in that case the pMAG continues to forward that to the pAN, and pAN
to the nAN to the MN.

It is a very smooth mechanism! :)


> 
> So it doesn't matter how you achieve this, the goal is to 
> enable the LMA to receive uplink packets from both the pMAG 
> and the nMAG for a certain amount of time during a handover.
> 
> Hope this clarifies my earlier email.
[Ahmad]
Thanks! it does. Please see comments above.

> 
> Vijay
> 
>