Re: [Gen-art] Gen-art review of draft-ietf-dime-pmip6-lr-07

[Elwyn Davies <[e-mail]>]

Hi, Qin.

Fine - I think all will be good from my point of view if you say a few
words about linking the MAGs as per your last response belwow.

Cheers,
Elwyn

On Wed, 2012-02-08 at 19:47 +0800, Qin Wu wrote:
> Hi, Elwyn:
> Thank for your followup comments, please see my replies inline.
> 
> Regards!
> -Qin
> ----- Original Message ----- 
> From: "Elwyn Davies" <[e-mail]>
> To: "Qin Wu" <[e-mail]>
> Cc: "General Area Review Team" <[e-mail]>; <[e-mail]>
> Sent: Wednesday, February 08, 2012 6:58 PM
> Subject: Re: [Gen-art] Gen-art review of draft-ietf-dime-pmip6-lr-07
> 
> 
> > Hi, Qin.
> > 
> > Thanks for your quick reponse.. some comments in line (I have deleted
> > the bits agreed),
> > 
> > /Elwyn
> > 
> > On Fri, 2012-02-03 at 12:53 +0800, Qin Wu wrote:
> >> Hi,Elwyn:
> >> Thank for your valuable review. please see our replies below.
> >> ----- Original Message ----- 
> >> From: "Elwyn Davies" <[e-mail]>
> >> To: "General Area Review Team" <[e-mail]>
> >> Cc: <[e-mail]>
> >> Sent: Thursday, February 02, 2012 9:06 PM
> >> Subject: Gen-art review of draft-ietf-dime-pmip6-lr-07
> >> 
> >> 
> >> >I am the assigned Gen-ART reviewer for this draft. For background on
> >> > Gen-ART, please see the FAQ at
> >> > < http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
> >> > 
> >> > Please wait for direction from your document shepherd
> >> > or AD before posting a new version of the draft.
> >> > 
> >> > Document: draft-ietf-dime-pmip6-lr-07.txt
> >> > Reviewer: Elwyn Davies
> >> > Review Date: 2 February 2012
> >> > IETF LC End Date: 24 January 2012
> >> > IESG Telechat date: 16 February 2012
> >> > 
> >> > Summary:
> >> > I have a couple of queries/minor issues regarding checking whether LMA1
> >> > and LMA2 are the same node and some hand waving over the idea of
> >> > 'localized routing setup/signaliing'.  There are also a few minor nits.
> >> > Otherwise this is ready for the IESG.
> >> > 
> >> > [This document missed the normal gen-art last call allocation
> >> > notification mechanism for some reason - so I didn't realize it was on
> >> > my allocation till the end of last call and as a result the review is a
> >> > bit late.]
> >> > 
> >> > Major issues:
> >> > None
> >> > 
> >> > Minor issues:
> >> > s5.1, para 3 and s5.2, last para:
> >> > In s5.1:
> >> >> MAG1 can verify
> >> >>    whether both MAGs are under the same LMA by comparing the addresses
> >> >>    of LMA1 and LMA2.
> >> > Is this guaranteed to work?  Should we care? Or is this just too bad if
> >> > the LMA has multiple addresses and the two MNs have different ideas?
> >> 
> >> [Qin]: In the example described in the s5.1, we don't consider one LMA has two LMA address.
> >> LMA1 and LMA2 may represent two different mobility entities identified by LMA1 adress and LMA2 adress respectively.
> >> If LMA1 address is LMA2 address, this just indicate LMA1 and LMA2 are the same mobility entity.
> >> Even one LMA have more than one LMA address, this still work since MAG can know if these LMA addresses come from
> >> the same mobility entity based on MN1 and MN2's binding update list.
> >> 
> >> > However in s5.2:
> >> >> In the case where MNs share the same LMA, LR should be initiated by
> >> >>    LMA1 (i.e.,LMA2) since only LMA1 knows that both MN1 and MN2 belong
> >> >>    to itself by looking up the binding cache entries corresponding to
> >> >>    MN1 and MN2. 
> >> > I am unsure whether these two statements are talking about the same
> >> > thing - and, if so, are they contradictory?
> >> > 
> >> 
> >> [Qin]: No Confliction, See above.
> > 
> > I think this is exactly the point: You give two different (but allegedly
> > non-conflicting ways) of doing the same thing at two places in the
> > draft.  From what you say, I infer that you could do either thing in
> > both cases. If so then it would be better to give the alternatives
> > together for the first case and refer to the previous comments when the
> > second case is reached in the text.
> 
> [Qin]: Good suggestion and will follow this.
> > 
> >> 
> >> > s5.1, last para:
> >> >> Figure 4 shows another example scenario, similar to the example
> >> >>    scenario illustrated in Figure 3, LMA1 does not respond to MAG1 with
> >> >>    the address of LMA2, instead setting up a localized routing path
> >> >>    directly between itself and LMA2 via localized routing signaling.
> >> > I am unsure what 'localized routing signaliing' would involve.  What
> >> > would the nodes do for this?  Appears to involve some waving of hands.
> >> 
> >> [Qin]: LMA1 and LMA2 exchange to trigger corresponding LMA to setup binding entries
> >> on the correspoding MAG for localized routing and establish localized
> >> routing path between MAG1 and MAG2.
> >> 
> >> > On a slightly broader point, there are a number of places where the
> >> > phrase 'localized routing setup' (or similar) is used.  It would, I
> >> > think, be useful to add a few words indicating what is thought to be
> >> > involved although actually doing it is clearly out of scope of this
> >> > document.
> >> 
> >> [Qin]: Okay.
> > 
> > I am afraid this doesn't really help: You say 'establish localized
> > routing path between MAG1 and MAG2'. How? Does this imply that the MAG
> > or some other component will (re-)configure the local packet routing
> > infrastructure? (Not something I would expect the MAG to be authorized
> > to do.) Or is this a matter of creating a tunnel?  I think this needs to
> > be a whole lot more concrete - both ends have to be expecting the
> > packets and know what to do with them.
> 
> [Qin]: Yes, your are right. Tunneling between MAG1 and MAG2 should be configured on both MAGs.
> As default static behavior,tunnel between MAG1 and MAG2 uses the same encapsulation mechanism
> as that being used for PMIP tunnel between MAG and LMA. In this case, MAG1 and MAG2
> can start using the same tunneling mechanism without special configuration (e.g.using other 
> tunnel mechanism that is uniform in the PMIP domain) on MAGs or dynamic tunneling negotiation between MAGs.
> but special configuration on MAGs or dynamic tunnel negotiation can overide 
> the default static behavior mentioned above if they are really needed.
> 
> Hope this clarifies.
> 
> > Regards,
> > Elwyn  
> > 
> >