Hi Tony,

> -----邮件原件-----
> 发件人: Tony Li [mailto:[email protected]] 
> 发送时间: 2009年8月5日 12:26
> 收件人: Xu Xiaohu
> 抄送: 'RJ Atkinson'; 'IRTF Routing RG'
> 主题: Re: [rrg] Some concerns about ILNP
> 
> 
> Let me see if I can answer:
> 
> > In fact, my question is much simple, that is: when a mobile 
> node moves 
> > to a new subnet where another node occasionally uses the same 
> > identifer as the mobile node, should the mobile node 
> renumber its identifier?
> 
> 
> Simply put: this needs to never happen.  That creates the 
> potential for a collision within the same locator.  It's up 
> to the administrator of that local subnet to ensure that 
> there's some mechanism for resolving this.  Note that it does 
> not mean that the mobile node much change its identifier, 
> just that it needs to not use that identifier while the other 
> node could possibly appear.
> 
> One alternate approach would be for it to have a temporary, 
> local identifier purely for use within that subnet.
> 
> Or, there can be some mechanism that prevents the other node 
> from connecting.

Do you mean if there is an identifiier collision during mobility, one should
be killed even though it is communicating with somebody else?
 
> > if so, can the
> > established session using that identifier survives after idenfifier 
> > renumbering?
> 
> 
> Certainly.  Again, the identifier does not need to be 
> abandoned.  The only restriction is that there are not two 
> systems with the same locator  and same identifier at the same time.
> 
> For example, if a node is going to make use of the temporary 
> identifier that I hallucinated above, then it could return to 
> its permanent identifier when it leaves the locator where 
> there's a conflict.  If the timing permits, the session can continue.

If a mobile node change the session association from its permenant id to its
temperary id, the session is broken.

> > if not, how could the last-hop router distinguishes these two hosts 
> > when forwarding packets destined to one of them?
> 
> 
> The last-hop router might well distinguish the two based on 
> the link layer address and knowing which one is active at present.

Before distinguishing the two identities, how could the last-hop router
determine which node a given link layer address is corresponding to? It seem
like a egg-chicken issue.

> > Could you please tell me in which I-D and which section the above 
> > question has been answered. Thus I can find the answer 
> quickly. Thanks.
> 
> 
> This is all in the base ILNP spec.

Thanks, I will read it carefully.

Best regards,

Xiaohu

_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to