RE: [MIP-QOS] RE: MIP-QOS -- Scope of discussions - One more comment

[Hemant.Chaskar@nokia.com]

> -----Original Message-----
> From: ext Vinod Kumar Choyi [mailto:vinod.choyi@usa.alcatel.com]
> Sent: 17. April 2001 20:19
> To: mip-qos@research.nokia.com
> Subject: Re: [MIP-QOS] RE: MIP-QOS -- Scope of discussions - One more
> comment
> 
> 
> How about provisions for, handoffs initiated by QoS requirements too.
> instead of only going by the usual, handoff to a cell on my 
> path, how about
> a detour to a cell that can offer me some kind of guarantees?

That is the igredient of a bigger problem called "service discovery" and is
outside the scope of our problem statement.

> How about soft reservations ?

I agree that reservations if any need to be soft state-based so that they
die away after handover. A requirement could be stated as;

"A QoS mechanism SHOULD release any QoS states created along the old
end-to-end path after handover."

> When do we decide to tear down reservations on segments that 
> are not part of
> the new path (immediately after handoff / after a certain 
> time period / what
> if we use Temporary Tunnels between Old and new ARs)
> What if I don't receive QoS support after a handoff can I use 
> Temporary
> tunnels between old and new Access Routers (Fast Handoff..) 
> to obtain the
> required QoS (It will be difficult because we would still be 
> using the same
> wireless link, but what if a segment above the AR cannot 
> provide me the
> support)
> 
> vinod kumar choyi
> 
> ----- Original Message -----
> From: <Hemant.Chaskar@nokia.com>
> To: <mip-qos@research.nokia.com>
> Sent: Tuesday, April 17, 2001 9:38 AM
> Subject: RE: [MIP-QOS] RE: MIP-QOS -- Scope of discussions - One more
> comment
> 
> 
> > After more careful look into the issue raised by Murray 
> Philip, now I
> > realize that there are a number of possible routes MN's 
> packets may take
> > after handover:
> >
> > 1. Non-optimized path which includes a segment from CN to 
> HA and another
> > from HA to MN.
> >
> > 2. Optimized path from CN to MN.
> >
> > 3. Temporary tunnel between old and new access routers 
> created by some
> > mobility protocols such as fast handover (note that even though the
> routers
> > may be physically adjacent, this tunnel could be a long 
> one, possibly even
> > crossing the core).
> >
> > Is it absolutely required to provide QoS support on all 
> these segment? Or
> > shall we just say that it is required to provide QoS 
> support on optimized
> > path and leave others as optimizations.?
> >
> > Hemant
>