Re: [OSPF] AD review of draft-ietf-ospf-prefix-link-attr-06

[Alia Atlas <[e-mail]>]

Hi Acee,

On Sat, Aug 1, 2015 at 5:54 PM, Acee Lindem (acee) <[e-mail]> wrote:

> Hi Alia,
> Thanks for the review.
>
> From: Alia Atlas <[e-mail]>
> Date: Thursday, July 30, 2015 at 2:22 PM
> To: "[e-mail]" <
> [e-mail]>, OSPF WG List <[e-mail]>,
> shraddha <[e-mail]>
> Subject: AD review of draft-ietf-ospf-prefix-link-attr-06
>
> As is customary, I have done my AD review
> of draft-ietf-ospf-prefix-link-attr-06 before asking for IETF Last Call.
> First, thank you very much for your hard work on this draft.  It is lovely
> to see needed work move quickly and have numerous interoperable
> implementations.
>
> I do have a number of minor issues on the draft - but all on the level of
> clarifications.  Therefore, I have requested that IETF Last Call be
> started.  Assuming good responsiveness on the part of the authors, a
> revised version that addresses my concerns can be on the IESG telechat on
> August 20.
>
> I do note that there are 6 authors on this draft.  Please provide input -
> since I know that you all well aware that the limit is normally at most 5.
> One can identify a primary editor or two.  This isn't pure process; the
> more authors listed on a draft, the longer it takes to handle AUTH48 -
> particularly when some are not as involved and do not respond rapidly and
> with full context.  I make no judgement about the authors of this draft -
> who have clearly moved from pulling out the idea into a stand-alone draft
> and had a number of different implementations.
>
>
> We’ve already made one pass on pruning the authors and since we are only
> one over, I’d like to leave it as is.
>

That's fine.  I hear that you have given it consideration and everyone is
actively participating and will not cause delays during AUTH48.

My review comments are below.   Thanks again for your hard work in getting
> this far!
>
> Minor issues:
>
> 1) On p. 6, it says " AF Address family for the prefix.  Currently, the
> only supported value is 0 for IPv4 unicast."  Please clarify VERY
> CLEARLY why this restriction exists.  Not everyone reading this will
> be familiar with support for IPv6 in various protocols and we are really
> finally heading towards lots more IPv6.
>
>
> I will clarify. There are basically two reasons. The first is that we
> really didn’t want to specify more than was necessary in this base
> document. The second is that we have OSPFv3 for IPv6. So, you may ask why
> we have this at all. The reason is that we didn’t want to rule out
> extension of OSPFv2 completely.
>

Well, it's easier to specify it for IPv6 & have it not used than to go back
and need to do another draft to add it later.  But this is a judgement call
on your part.  Thanks for clarifying the reasoning in the draft.

2) On p. 6 and 8.:
> "The Instance field is an arbitrary value used to maintain multiple
> Extended Prefix Opaque LSAs.  A maximum of 16777216 Extended Prefix
> Opaque LSAs may be sourced by a single OSPF instance.": This doesn't
> really give normative behavior.  I assume that what you mean is that
> the advertising router has a number space for the Instance which has
> no significance outside of that advertising router and can have
> arbitrary values allocated from it.  Each of these LSAs is identified
> uniquely by its Instance number.  Please provide good text for what
> MUST be done and indicate that the value may be used for tie-breaking
> ("In this case, the Extended-Prefix-TLV in the Extended Prefix Opaque
> LSA with the smallest Instance is used by receiving OSPFv2 Routers. ")
> and there's an assumption that the values will be allocated from
> smallest to largest.
>
>
> I will clarify this. However, I don’t want to specify any assumption about
> allocation.
>

I wasn't trying to put in assumptions about the allocation, but if you read
further in
the draft, there are assumptions in the tie-breaking based upon lower
instance values
being allocated first.  Consistency is preferred. :-)


> 3) On p. 6 for the Route Type, it would be useful to have a reference to
> where these type values are pulled from.  I'd also like to see some
> text about whether other values could be valid in the future and how
> so.  For instance, I'm assuming that you are basically pulling the
> values from the OSPFv2 Link State (LS) Type
> (
> http://www.iana.org/assignments/ospfv2-parameters/ospfv2-parameters.xhtml#ospfv2-parameters-5
> )
> - so perhaps you could simply say so or clarify for what are valid
> values.
>
>
> Is there an example of referencing an IANA directory? I could also
> reference RFC 2823 and RFC 3101 directly.
>

I was just thinking of referencing the name of the IANA registry; having
the URL would be fine in the reference
section, I think. If not, I'm sure the RFC Editor will let us all know ;-)


4) On p. 9: For Link-Type, could you also put a reference to the IANA
> registry?  I'd prefer it to be clear that if (unlikely as it seems)
> there were a new Link-Type added, it would apply here too.
>
>
> Sure. We’ll do the same.
>
>
>
> 5) In Sec 5, pleaes add an RFC Editor note that Section 5 will be removed
> upon publication.  That's the intent wtih RFC 6982.  Thanks for
> including this section in the draft.  If the information wants to move
> to the OSPF WG wiki, that would give it a place to survive after this
> draft is submitted to the RFC Editor.
>
>
> Ok - I need to hunt down this OSPF Wiki we talked about in Prague as well.
>

The IDR one is at http://trac.tools.ietf.org/wg/idr/trac/wiki.  The OSPF
one would
be http://trac.tools.ietf.org/wg/ospf/trac/wiki but it looks like it isn't
there.  You may
need to contact the tools team ([e-mail] or more usefully
[e-mail]).

Once it is created, I can also add it to the charter page on the
datatracker - along with
any other URLs that might be useful.

Thanks,
Alia

>
> Nits:
>
> 6) In Sec 2, there's an "e.g., mapping server deployment".  Could you add
> a reference?  This tells me nothing...
>
>
> Sure - it is the segment routing architecture document.
>




>
>
> 7) In Sec 2, In the packet format, could you clarify Opaque type = 7? Same
> for on p.8 for opaque type = 8 ?
>
>
> Sure.
>
>
> 8) Since you are creating the registry for the TLVs, please clearly state
> that value 1 is being used earlier - instead of "suggested value" as
> on p.9
>
>
> Sure.
>
> Thanks,
> Acee
>
>
>
> Regards,
> Alia
>
>