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

Hello Nicolas

> Hi Ryuji,
> 
> Ryuji Wakikawa wrote:
> 
> >Hi
> >
> >  
> >
> >>>-----Urspr|ngliche Nachricht-----
> >>>Von: [email protected] [mailto:owner-mobile-
> >>>[email protected]] Im Auftrag von Ryuji Wakikawa
> >>>Gesendet: Dienstag, 22. Juli 2003 12:18
> >>>An: [email protected]
> >>>Cc: [email protected]; [email protected]; mobile-
> >>>[email protected]
> >>>Betreff: Re: [nemo] WG: [mobile-ip] new I-D on HA filtering
> >>>
> >>>Hello,
> >>>
> >>>At Wed, 23 Jul 2003 15:11:36 +0200,
> >>>Thomas Noel wrote:
> >>>      
> >>>
> >>>>Hi Koojana, All,
> >>>>        
> >>>>
> >>>>>Perhaps, we should initiate a discussion in mobile-ip or nemo WG to
> >>>>>          
> >>>>>
> >>>discuss
> >>>      
> >>>
> >>>>>about this topic.
> >>>>>
> >>>>>          
> >>>>>
> >>>>I Agree with you, we have now several drafts and it will be interesting
> >>>>to converge towards an unique solution to solve this issue.
> >>>>        
> >>>>
> >>>Can we discuss filters/policies and multipleCoAs registration separately?
> >>>      
> >>>
> >>IMHO, Multiple CoAs and Flow Distributions should be discussed together. The
> >>reason is because once multiple CoAs are available, the natural requirement 
> >>that follows would be to use these points of attachments efficiently.
> >>    
> >>
> >
> >I understand there is a way to manage filters at the IP level, but
> >there are also other ways to manage them. If a user wants to manage
> >filters information statically, does she/he still need to support
> >filters management at IP layer? 
> >
> What way do you think to manage filters ? If you perform it at uppers 
> layers, you have to treat the packets by the lower layers, haven't you ?

Mobile IP need to parse packets to decide the outgoing interface somehow.
MIP accesses some filters database to decide an appropriate interface or
CoA, but "management of the database is not MIP oriented operations." 

In your model, application or users configures filters to MobileIPv6 and
MIP notifies the filters to HA and CN. 

I don't want to carry filters only by MobileIPv6. 
For example, user can configure filters to the database manually before
communication. or an application may configure these by itself. 
If each interfaces is highly abstracted, anybody can configure 
filters to the database. 

If you need an example, I will explain more.

> Performing filtering at the IP layer in MIPv6 allows a MN to request 
> your home agent to not even forward potential packets that may reach 
> your home network (rather than load your interface with traffic you 
> don't want to).
>
> Or for the same CN, to send two flows to different CoAs associated with 
> the same home address.

I think my wording is incorrect. sorry.
I wanted to say "I don't like carrying filters information by using MIP
signaling only".

I am OK if there is a way to carry filters by MIP signaling, but we
should allow to support the other ways. This is why I asked you to
separate discussion.  I never ask you to stop discussing filter stuff:-)


regards,
ryuji


> regards,
> Nicolas




> 
> >
> >regards
> >ryuji
> >
> >
> >
> >  
> >
> >>>Most of draft needs multiple CoAs registrations, but some of systems
> >>>do not want to support filters/policies on the IP layer.
> >>>
> >>>Our draft(multiplecoa) deals with only multiple CoAs registration.
> >>>
> >>>      
> >>>
> >>>>Perhaps, Nemo is the ideal WG to discuss about this topic.
> >>>>        
> >>>>
> >>>It might be better to keep discussing it in MIP6/MIPSHOP WG, because
> >>>NEMO is based on Mobile IPv6. Once MIP supports multiplecoa, filters
> >>>whatever, NEMO can have the same feature.
> >>>      
> >>>
> >>As Hesham suggested, I too believe that mip6 or mipshop would be the best 
> >>place to discuss as it should be a extension to the MIPv6 base
> >>specification, which is not considered simultaneous bindings as in the MIPv4
> >>RFC.
> >>
> >>Kind Regards
> >>Koojana
> >>
> >>    
> >>
> >>>If NEMO still needs further operations, then NEMO becomes the ideal WG.
> >>>
> >>>best regards,
> >>>ryuji
> >>>
> >>>      
> >>>
> >>>>Regards
> >>>>
> >>>>Thomas Noel
> >>>>LSIIT
> >>>>
> >>>>        
> >>>>
> >>>>>Kind regards
> >>>>>
> >>>>>Koojana
> >>>>>
> >>>>>-----Urspr|ngliche Nachricht-----
> >>>>>Von: [email protected]
> >>>>>[mailto:[email protected]] Im Auftrag von Nicolas
> >>>>>Montavont
> >>>>>Gesendet: Mittwoch, 23. Juli 2003 09:55
> >>>>>An: mobileip
> >>>>>Betreff: [mobile-ip] new I-D on HA filtering
> >>>>>
> >>>>>Hi all,
> >>>>>
> >>>>>We just submitted a new I-D on Home Agent Filtering for Mobile IPv6,
> >>>>>available at
> >>>>>
> >>>>>http://www-r2.u-strasbg.fr/~montavont/draft-montavont-mobileip-ha-
> >>>>>          
> >>>>>
> >>>filtering-
> >>>      
> >>>
> >>>>>v6-00.txt
> >>>>>
> >>>>>It will be soon available in the IETF repertories.
> >>>>>
> >>>>>This I-D proposes a solution for the MN to set filters on its home
> >>>>>agent. Filters can be used to spread flows on different CoAs (and
> >>>>>potentially several network interfaces), and to forbid the redirection
> >>>>>of certain flows from the home agent.
> >>>>>
> >>>>>Here is the abstract:
> >>>>>
> >>>>>Mobile IPv6 allows a MN to receive incoming packets to its home
> >>>>>          
> >>>>>
> >>>address
> >>>      
> >>>
> >>>>>while it is away from its home network. In a heterogeneous
> >>>>>          
> >>>>>
> >>>environment,
> >>>      
> >>>
> >>>>>a MN may have multiple interfaces, each with different
> >>>>>          
> >>>>>
> >>>characteristics.
> >>>      
> >>>
> >>>>>While a MN is in a visited network, due to the performance of the
> >>>>>interfaces or to the user preferences, the MN may want to forbid the
> >>>>>redirection from its home agent of a kind of flow, or to indicate a
> >>>>>target CoA for a kind of flow.
> >>>>>In this draft, we propose new mobility options that allow a MN to
> >>>>>advertise filters to its home agent. A filter is associated with a
> >>>>>          
> >>>>>
> >>>CoA,
> >>>      
> >>>
> >>>>>in such a way that the MN can register several CoA and can register
> >>>>>several filters for one CoA. A filter may indicate that a flow which
> >>>>>          
> >>>>>
> >>>map
> >>>      
> >>>
> >>>>>to a filter must be dropped or must be redirected to the indicated
> >>>>>          
> >>>>>
> >>>CoA.
> >>>      
> >>>
> >>>>>Regards,
> >>>>>Nicolas
> >>>>>
> >>>>>
> >>>>>
> >>>>>          
> >>>>>
> >>>>
> >>>>        
> >>>>
> >
> >  
> >
> 

regards,
ryuji