[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [netlmm] I-D Action:draft-ietf-netlmm-lma-discovery-01.txt



Hi,
  
  MAG making a DNS query for a service name and randomly selecting one LMA from among the returned replies is a good way of discovering LMA for each MN:
- it does not require MN providing input so it works with unmodified hosts,
- LMA load balancing is achieved.

Regards,

Behcet


----- Original Message ----
> From: jouni korhonen <jouni.nospam at gmail.com>
> To: Xiangsong Cui <Xiangsong.Cui at huawei.com>
> Cc: netlmm at ietf.org
> Sent: Thursday, August 27, 2009 8:12:38 AM
> Subject: Re: [netlmm] I-D Action:draft-ietf-netlmm-lma-discovery-01.txt
> 
> Hi,
> 
> On Aug 27, 2009, at 11:57 AM, Xiangsong Cui wrote:
> 
> > Hi Jouni,
> > 
> > Let's take a precise analyse on this issue.
> > 
> > You said such information may come from lower layers,
> > but who generate this information?
> > 
> > Does lower layer generate the LMA information or the
> > LMA information does depend on the link type?
> > 
> > I think it is not.
> > 
> > Maybe it is the user who configures the LMA information
> > in the mobile node (i.e. the device), but in the MAG perspective,
> > the initiator of the message which contains the LMA information,
> > the initiator is the MN, the MAG gets the LMA information from
> > the mobile node.
> > 
> > For example, the MAG gets LMA information from AAA server,
> > and the LMA information in AAA server is configured by the administrator.
> > We can say MAG gets LMA information from AAA server but nobody
> > says MAG gets LMA information from the administrator.
> > 
> > So I think it is a MN-assist solution.
> 
> The MN is only doing what it would do independent of PMIP being deployed or not 
> and the MAG just makes advantage of it. Some existing systems & MNs already do 
> this.
> 
> Based on the discussion we have had I think it is now worth keeping this 
> solution in the I-D. I would, however, reword it like:
> 
>   The solution described in this section is similar to the solution
>   discussed in Section 3.1.  Instead of deriving the LMA FQDN from the
>   MN identity, the MAG receives explicit LMA FQDN or IP address
>   information from lower layers, for example, as a part of the normal
>   lower layer signaling when the MN attaches to the network.  This
>   usually means the MN is also the originator of the LMA information.
>   However, the LMA information content as such can be transparent to
>   the MN, meaning the MN has no knowledge it being anything LMA
>   related.
> 
> 
> Cheers,
>     Jouni
> 
> 
> 
> > 
> > 
> > Regards/Xiangsong
> > 
> > 
> > ----- Original Message ----- From: "jouni korhonen" 
> > To: "Xiangsong Cui" 
> > Cc: "Vijay Devarapalli" ; 
> > Sent: Thursday, August 27, 2009 4:15 PM
> > Subject: Re: [netlmm] I-D Action:draft-ietf-netlmm-lma-discovery-01.txt
> > 
> > 
> >> Hi,
> >> 
> >> On Aug 27, 2009, at 5:11 AM, Xiangsong Cui wrote:
> >> 
> >>> Hi Vijay,
> >>> 
> >>> I think I got your point, but I still have some questions.
> >>> 
> >>> 1, NETLMM WG is about mobility management, so the item
> >>>  of LMA discovery is of course in scope of mobility management,
> >>>  is it right?
> >> 
> >> ok.
> >> 
> >>> 
> >>> 2, MN provides LMA information to MAG, is a *MN-assistant*
> >>>  LMA discovery, is it right?
> >> 
> >> We are a bit on a grey area here. Like the text says such information  may 
> come from lower layers. If such information delivery is e.g. part  of the normal 
> L2 attach procedure independent whether there is PMIP or  not, I would not call 
> it in that case MN assisted LMA discovery.
> >> 
> >>> 
> >>> 3, MN-assistant LMA discovery means MN involves in the mobility
> >>>  management, is it right?
> >> 
> >> Continuing from the above. If the information that MN provides is configured 
> into MNs and this configuration is independent whether  there is PMIP or not, I 
> would not say that MN is involved in the  mobility management. The content of 
> the configuration data might then  be different depending whether PMIP is used 
> but that is transparent to  MNs.
> >> 
> >> Cheers,
> >> Jouni
> >> 
> >> 
> >>> 
> >>> Regards
> >>> 
> >>> Xiangsong
> >>> 
> >>> 
> >>> ----- Original Message ----- From: "Vijay Devarapalli" > >>> >
> >>> To: "Xiangsong Cui" ; "jouni korhonen" 
> > >>> >
> >>> Cc: 
> >>> Sent: Thursday, August 27, 2009 8:32 AM
> >>> Subject: RE: [netlmm] I-D Action:draft-ietf-netlmm-lma- discovery-01.txt
> >>> 
> >>> 
> >>> You are right. I am not disagreeing that the mobile node knows that
> >>> PMIPv6 is being used in the network when it sends the LMA  information to
> >>> the MAG. But we have a sort of working assumption that the MN is  able to
> >>> supply certain information to the MAG at L2, for example, attach  versus
> >>> handover indication. This would be similar to that.
> >>> 
> >>> Vijay
> >>> 
> >>>> -----Original Message-----
> >>>> From: Xiangsong Cui [mailto:Xiangsong.Cui at huawei.com]
> >>>> Sent: Monday, August 24, 2009 7:13 PM
> >>>> To: Vijay Devarapalli; jouni korhonen
> >>>> Cc: netlmm at ietf.org
> >>>> Subject: Re: [netlmm] I-D
> >>>> Action:draft-ietf-netlmm-lma-discovery-01.txt
> >>>> 
> >>>> Hi Vijay,
> >>>> 
> >>>> In my understanding, LMA is a functional element of mobility
> >>>> management, the mobile node with only simple IP stack can't
> >>>> recognize the information of LMA and can't transmit it to MAG.
> >>>> 
> >>>> If the mobile node can't send this information to MAG,
> >>>> MAG can't receive this information from mobile node.
> >>>> 
> >>>> Regards/Xiangsong
> >>>> 
> >>>> 
> >>>> ----- Original Message ----- From: "Vijay Devarapalli" > >>>> >
> >>>> To: "jouni korhonen" ; "Xiangsong Cui"
> >>>> 
> >>>> Cc: 
> >>>> Sent: Tuesday, August 25, 2009 1:47 AM
> >>>> Subject: Re: [netlmm] I-D
> >>>> Action:draft-ietf-netlmm-lma-discovery-01.txt
> >>>> 
> >>>> 
> >>>> Hello,
> >>>> 
> >>>> On 8/24/09 1:56 AM, "jouni korhonen" wrote:
> >>>> 
> >>>> >> Section 3.2,
> >>>> >> "  This usually means the MN is the
> >>>> >>  originator of the LMA information and explicitly
> >>>> participates to the
> >>>> >>  mobility management signaling "
> >>>> >>
> >>>> >> In NETLMM, the principle is that the MN should not be involved in
> >>>> >> mobility management, right?
> >>>> >> So the solution "Receiving LMA FQDN or IP Address" should be
> >>>> >> deleted in the netlmm WG draft.
> >>>> >
> >>>> > I would be fine deleting this section. Others?
> >>>> 
> >>>> I disagree. The section explicitly says the MAG receives the
> >>>> LMA FQDN or IP
> >>>> address from the MN in the lower layers. This is not the same
> >>>> as the MN
> >>>> doing mobility management.
> >>>> 
> >>>> Vijay
> >>>> 
> >>>> 
> >>> 
> > 
> 
> _______________________________________________
> netlmm mailing list
> netlmm at ietf.org
> https://www.ietf.org/mailman/listinfo/netlmm