Re: [multimob] Comments on draft-deng-multimob-pmip6-requirement-01.txt

Hi Behcet,

Behcet Sarikaya wrote:

> [thomas] RFC5213 does not address multicast routing (that's why we are
> discussing it). The scenario described in fig 2 is *not* addressing
> mobile multicast senders,
> 
> [behcet] Then I think the text and figure on the local routing is not 
> needed because local routing in PMIPv6 refers to the routing of packets 
> coming from the uplink and as you said this would be multicast sender 
> situation which you left out in the requirements draft.
>

[thomas] This is a misunderstanding: Fig 2 does *not* refer to sender 
mobility.

> but a typical content distribution setting as
> for instance used in a provider-centric IPTV service. This means, a
> provider issues multicast streams from within his network domain, aiming
> at an efficient way to preserve (his own) network resources.
> 
>  From the minimal perspective of RFC5213 (unicast routing) there is no
> efficient way to do this. That's why we are trying to identify protocol
> extensions that smoothly cooperate with RFC 5213.
> 
> [behcet] The situation you are talking about could become valid if route 
> optimization is used with PMIPv6. In that case the content server in 
> between MAG and LMA would send directly to MAG. If route optimization is 
> not used then all traffic goes through LMA-MAG tunnel. Route 
> optimization for PMIPv6 is some future work, there are drafts but we 
> don't know if the standardization will go ahead.
> Route optimization with multicasting is even more complicated problem. 
> If you wish to state some requirements for this very complicated case 
> please go ahead.

[thomas] Hmm, I'm not trying to add complicated requirements. This is 
the overview about possible scenarios.
  We're all just trying to think of a reasonable scope to address and 
solve the problem (so does the recent pmip6-extension-draft ... see also 
the last mail from Hitoshi).

If multicast on PMIPv6 is supposed to serve as a solution for 
large-scale content distribution (like in IPTV applications), then we 
should not just propose to tunnel all traffic from one LMA to the MAGs. 
That's a bitter pill for those who have to pay for the infrastructure.

Best regards,

Thomas

-- 

° Prof. Dr. Thomas C. Schmidt
° HAW Hamburg, Dept. Informatik
° University of Applied Sciences
° Berliner Tor 7, D 20099 Hamburg, Germany
° Fon: +49-40-42875-8452, Fax: -8409
° http://www.informatik.haw-hamburg.de/~schmidt
_______________________________________________
multimob mailing list
multimob@ietf.org
https://www.ietf.org/mailman/listinfo/multimob