Re: [Hipsec] HIP Registration problem

[Andrew McGregor <[e-mail]>]

Register with several and use DNS round-robin, or just talk to all of them at once.  That's probably good enough, I think.

Andrew

On 8/11/2010, at 11:03 AM, Zhangfeng HU wrote:

> Hi,
> yes, that's the truth. Since HIP does provide a mechanism as DAAP of MIP, we are just considering a method of providing fault torlerance for RVS in HIP. For some mobile nodes, the high availability is very important, however the availability of these mobile nodes is mostly decided by the availability of the corresponding RVS who serves them. This RVS fault torlerance can not be easily implemented through updating the entries in DNS, because of the high latency and the cache mechanism used in DNS.
>  
>  
>  
> 2010-11-08
>  
> Zhangfeng HU, PhD candidate of
> Broadband Network Research Center,
> State Key Laboratory of Networking and Switching Technology,
> Beijing University of Posts and Telecommunications, Beijing, China
>  
> Email:[e-mail]
> Tel:86+13811892137
> 发件人： Miika Komu
> 发送时间： 2010-11-04  18:47:45
> 收件人： hipsec
> 抄送：
> 主题： Re: [Hipsec] HIP Registration problem
> On 11/04/2010 11:48 AM, zfhu2001 wrote:
> > Hi all,
> > As we know, a mobile node registers its HIP and IP address binding to a 
> > correponding RVS server, but now I'm wondering whether the MN will 
> > register to a new RVS if a mobile node moves to another network and 
> > simultaneously its previous RVS server does not work any more. Or does 
> > HIP provide a mechanisim as DAAP of MIP to detect the serving RVS server 
> > of its home area?
> > Thanks.
> > 2010-11-04
> Hi,
> there's no such thing as "home agent" or "home area" in HIP. This is one
> of the differences to MIP. The same rendezvous server can be reused
> independently of the current network using the UPDATE procedure.
> Remember also that the rendezvous server forwards only the first I1 or
> the first UPDATE packet, so it's still a light-weight operation even
> though the latency be longer. The rendezvous server does not forward any
> application payload.
> Fault-tolerance for RVS may fall into the category of local policies
> (?). Changing the "current" RVS requires also an update to the DNS
> records. Another strategy could be just to publish multiple RVS servers
> in the DNS and send packets to all of them (which ever happens to be alive).
> _______________________________________________
> Hipsec mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/hipsec
> ÔÚ´«ÈëµÄÓʼþÖÐδ·¢ÏÖ²¡¶¾¡£
> ¼ì²é¹¤¾ß£ºAVG - www.avg.com 
> °æ±¾£º9.0.865 / ²¡¶¾Êý¾Ý¿â£º271.1.1/3235 - ·¢²¼ÈÕÆÚ£º11/03/10 16:36:00
> _______________________________________________
> Hipsec mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/hipsec