RE: [Mip6] IETF 58 Meeting minutes

["Maeva Petitjean" <[e-mail]>]

Dear Gopal,

Erratum

*Please Read*
 http://www.mindspring.com/~bpatil/IETF58/MIP6

*and not*
> MIP6 WG
> Monday November 10th, 2003
> Chairs: Basavaraj Patil, Gopal Dommety
> http://www.mindsrping.com/~bpatil/IETF58/MIP6

Thank you
Regards,

Maeva PETITJEAN
ETSI Plugtests

> -----Original Message-----
> From: Gopal Dommety [mailto:[e-mail]]
> Sent: 08 December 2003 19:07
> To: [e-mail]
> Subject: [Mip6] IETF 58 Meeting minutes
> 
> 
> Hello,
> 
> I am attaching the meeting min for MIP6 WG. Thanks to Eva 
> Gustafsson and 
> Alepsh Patel for taking the meeting notes.
> 
> Regards
> Gopal
> 
> 
> 
> MIP6 WG
> Monday November 10th, 2003
> Chairs: Basavaraj Patil, Gopal Dommety
> http://www.mindsrping.com/~bpatil/IETF58/MIP6
> 
> Agenda Bashing (Basavaraj Patil)
> Document status (Basavaraj Patil)
> 
> MIPv6 docs to be moved to the MIP6 WG from the Mobile IP WG 
> (which is closing)
> IANA assignments for the base spec ongoing
> 
> MIPv6 remote interop testing (Samita Chakrabarti)
> 
> Connectathon 2004 was announced
> mobileip-ipv6, mipv6-ha-ipsec, perhaps nemo basic support 
> draft will be tested
> Connectathon update - new co-ordinator (UNH - Hiroshi T)
> 
> Design team was formed to develop guidelines  for uniform 
> remote testing 
> over IPv6 Internet.
> This effort is similar  to 6-bone testing. This effort does 
> not replace 
> regular MIPv6 interop events
> First draft is an individual submission. Need to get WG 
> consensus if this 
> should be a WG doc, on the interest and the Informational 
> nature of the draft.
> 
> The purpose of remote interop testing is for  some 
> implementers to dedicate 
> stable MIPv6 systems in the Internet, other implementers can 
> check basic 
> functionality.
> 
> Requirements to participate:
>   all nodes must register through ETSI web page, dedicated 
> HAs, MNs & CNs 
> in the test network
> Central registration at: list.etsi.org/plugtests-mip6.html
> 
> Comment from someone (could not catch the name) about helping 
> out with 
> reviewing the test plan.
> 
> Continuing... (TJ Kniveton)
> 
> Discussed issues of IPv6 address allocation, security 
> association, virtual 
> home link,
> diagram of nodes, solicited more implementer's comments are welcome
> 
> Current thought is to support Security associations - three 
> methods to auth 
> MN: NONE, AH, ESP
> 
> 
> Inter-home agents protocol (HAHA) (Ryuji Wakikawa)
> 
> Discussed the Problem statement:
>   need for reliability,
>   load balancing (HA can be serving large number of MNs,
>   if MN not uses RO HA becomes bottleneck of communication;
>   several HAs sharing the load for the same home network),
> redundancy (with HAs on topologically different links), MN cannot
> currently register its binding to multiple HAs simultaneously,
> serviceability, HA switching.
> 
> Discussions on mailing list: applying existing protocols 
> (VRRP/HSRP, BGP
> model, CARP-like mechanisms)
> 
> Goals are redundancy, load sharing, flexible HA selection
> 
> James Kempf: Want to do ipsec between HAs? (Yes) If you use ICMP you
> can't because the spec prevents you from doing it.
> 
> TJ Kniveton: MN can change to be registered with best HA, how do you
> figure out which is best?
> 
> Ryuji: MN knows some of the HAs...
> 
> TJ Kniveton: Needs some more thinking... MN home address will be moved
> to new HA. What if HA advertises address that is not topologically
> correct...
> 
> Q: If you route packets back to primary HA, how do you then 
> reduce load
> on HA? This actually increases load on HA. Also, if two HA 
> advertise the
> same prefix MN will only choose one. How will second prefix enter
> routing table? Discussion about this ending with "Don't think 
> you get my 
> question..."
> 
> Q: Big aggregated prefix will be split into several... to reduce load
> you don't route traffic back...
> 
> Basavaraj Patil: Need to discuss what the problem is that the WG wants
> to work on. Need problem statement & problem definition. Then decide
> this is problem really relevant that we should work on. This is one
> solution, there are probably more. We need to look at that.
> 
> Alpesh Patel: Routing with multiple anycast addresses popping 
> up...? Any
> analysis on this?
> 
> Ryuji: Will make problem statement & send to the list.
> 
> 
> MIPv6 advanced socket API extensions (Samita Chakrabarti)
> 
> API is useful for debugging, tracing, policy applications etc.
> This is a combined effort of IPv6 & MIP6 WG
> 
> Updates & resolved issues, suggested changes.
> 
> Next steps : Basavaraj Patil: Discussions with IPv6 WG, happy 
> to let work 
> be done
> here, in conjunction with IPv6 WG, ok to make this a WG document if
> there is WG consensus.
> 
> 
> Problem statement for multi-homed mobile nodes (Nicolas Montavont)
> 
> List of current work on multi-homing
> What's missing: terminology, motivation, accurate & common goal
> Some issues are related to Mobile IP while some are not
> 
> A MN is multi-homed when it has several addresses to choose between
> Benefits of multi-homing: redundancy, ubiquitous computing, load
> balancing, preference settings
> Why MIPv6?
> Open issues
> Which WG to take this to?
> 
> James Kempf: This is what I want to talk about at bar BoF 
> tonight. Have
> invited people from transport area to come.
> 
> Q: Some issues related to MIPv6, some not, those related to MIPv6 are
> obviously for this WG.
> 
> Basavaraj Patil: Aspects for MIPv6 are interesting. Making bigger
> changes to current MIPv6. How important is it to get this done right
> now? Need to also look at if problem can be solved in different way.
> Suggestion to look at problem statement & scenarios.
> 
> Thierry Ernst: Good to investigate problem statement, but 
> where do we do
> it? This WG? Nemo WG? How to coordinate activities between WGs.
> 
> Q: Node with multiple interfaces... that's the problem.
> 
> Raj: ADs have any comments?
> 
> Thomas Narten: No clear answer. Clarify problem statement, and ask
> question about which WG for each specific problem. Maybe 
> there is no WG,
> maybe we need BoF.
> 
> Basavaraj Patil: This is problem also seen in nemo. Don't start new WG
> and work on generic solution. Have to consider fact that 
> MIPv6-specific
> issues belong in this WG.
> 
> Q: Some issues are very specific; which interface node picks 
> is totally
> internal to node. How multiple pieces are put together? Applicable to
> MIPv6: you cannot have MN with multiple addresses work with 
> MIPv6 today.
> 
> Basavaraj Patil: Let's continue discussion, look at 
> scenarios, see what
> comes out of bar BoF.
> 
> 
> MIPv6 MIB (Glen)
> 
> Purpose: monitoring operations of mip6 entities (MN, HA, CN);
> operational statistics, errors, events
> Configuring/controling mip6 entities
> 
> MIB design
> 
> Greg Daley: Don't have to bind same home address to same c/o 
> address in
> all cases... CN1 & CN2, you want to communicate with them on two
> different paths, you want to have entry in table... multiple 
> c/o address
> to one home address, BU list tells you who you are communicating with;
> you have no information about that in this table
> 
> Samita Chakrabarti: What are address types
> 
> Glen: Here type is address type
> 
> Greg Daley: Does registered also mean re-registered?
> 
> Glen: No
> 
> Q: When you come back home you can de-register. This does not expire.
> Want to make that distinction?
> 
> Updated I-D, MIB issues: MIB review guidelines, RFC issues: security
> considerations
> Related issues, to be done, implementation
> 
> 
> Route optimization hint option (Keiichi Shima)
> 
> Problem statement: MIP6 defines route optimization mechanism, 
> there are no 
> guidelines when
> to start route optimization
> Things to be discussed & defined: condition on when to use RO, methods
> to inform MN about condition.
> 
> Example of configuration: fire-walled network
> 
> Q: why not just configure firewall to let packets through? Firewall
> never blocks response to request. Mobility header. Don't 
> understand what
> problem you are trying to solve here.
> 
> Basavaraj Patil: Yes, this is firewall problem. Let's take this
> off-line.
> 
> Jari Arkko: Would be better to update firewall. Also, wouldn't it be
> better to let message get dropped and inform MN there is no RO (if
> firewall blocks). That's the way the specification currently works.
> 
> Greg Daley: Other issues than firewall cases? What about 
> latency between
> MN & CN? You reduce signaling each time you move if you don't 
> have to do
> RO.
> 
> Basavaraj Patil: This draft only looks at one instance where RO should
> be used or not. There are lot more cases, and therefore lot more work
> that needs to be done on this.
> 
> Q: Valid problem, but depends whether 
> organization/corporation wants to
> go to standardization or just reconfigure / solve internally.
> 
> 
> Roadmap for MIP6 (Gopal Dommety)
> 
> Home agent reliability
> Alternate RO schemes
> MIP for MIP6
> Bootstrap for SA between MN & HA
> Document issues
> Separate specs for issues such as....
> 
> Charlie Perkins: Combine the charter of this WG with mipshop?
> 
> Gopal Dommety: Out of scope of the roadmap discussion.
> 
> Q: What are we planning to do with bootstrap? Where is 
> mechanism coming
> for bootstrap.
> 
> Basavaraj Patil: We had lengthy discussion in Vienna.
> 
> Jari Arkko: One thing missing is if you add new home address, 
> setting up
> policy entrance is not possible.
> 
> Discussion items (Gopal Dommety)
> 
> Identity for authentication (NAI)
> Multi-homing discussion
> Deployment issues discussion
> Vendor-specific extensions
> Authentication of MN to HA without ipsec
> 
> Looking for people interested in working on bootstrap mechanisms.
> 
> 
> 
> Acknowledgements:
> 
> Thanks to Eva   Gustafsson and Alpesh Patel for taking 
> excellent meeting 
> notes.
> 
> 
> 
> 
> _______________________________________________
> Mip6 mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mip6
> 

_______________________________________________
Mip6 mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/mip6