AW: AW: AW: [nemo] WG: new I-D on HA filtering

Hi Ken
 
> >> I think we discuss filters/policies and multiple CoAs registration
> >> separately. Because they are independent and separate discussion is
> >> more simple.
> >>
 
> K> This aspect is true. But, when there are multiple CoAs, flow
> distribution
> K> should naturally be present, instead of using one CoA or multiplying
> packets
> K> to each CoA.
 
 
> K> Our focus (as defined in
> K> http://www.ietf.org/internet-drafts/draft-nomadv6-mobileip-filters-
> 00.txt)
> K> is to keep multiple CoA registrations without duplicating traffic to
> each
> K> active Point of Attachments. For this, we use a flow distribution
> mechanism.
> 
> K> I believe, multiple CoA registration & filters/policies go together.
> 
> I would like to know why you believe they go together?


If we consider multiple CoAs, we SHOULD think how HA or CN forward traffic
to different CoAs of MN. Then, we can apply filters/policies for this
purpose. For ex, MIPv4 considers simultaneous bindings, but no body (as far
as I know) uses that concept (Enabling s bit in Registration Request), since
it leads to duplicate traffic to all active point of attachment.