Alvaro

Thanks. Please see inline.

On 22/09/2014 23:55, Alvaro Retana (aretana) wrote:
> Hi!
>
> After a long wait, I finally got to this..
>
> I do have some comments (minor and nits) that I would like addressed.
> As you look at these I will get the write up ready to send to the IESG
> --- you can either address my comments now/before the IESG gets the
> publication request or with any comments that come from them.
>
> Thanks!
>
> Alvaro.
>
>
> Minor Issues
>
>  1. 4.2.1.3 states that "In Figure 1 the intersection of the E's
>     Q-space with S's P-space defines the set of viable repair tunnel
>     end-points. . .there is no common node and hence no viable repair
>     tunnel end-point." The immediately following section (4.2.2)
>     starts with "The mechanisms described above will identify all the
>     possible repair tunnel end points that can be used to protect a
>     particular link." So..you're saying that there are no viable
>     end-points (which is technically true because you're using the
>     P-space)..and then you say that you figured out all the possible
>     repair points, which is obviously not true because in the previous
>     section you didn't mention the union of the Q-space and the
>     extended-P-space. All this is to say that a reader may be confused
>     --- adding mention of the extended p-space in 4.2.1.3 should clear
>     it up.
>
I added some text to address this.
>
>  1. Section 5 (Example). "When a failure occurs on the link between
>     PE1 and P2", but the figure doesn't show these 2 routers as
>     connected..so I'm assuming you're talking about PE1 and P1 (and
>     later in the text PE2 and P2).
>
I think there were a couple of errors here. Please can people take a
look at -7 and makes sure I have not made any other silly errors.
>
>  1. Section 6. What is the "null strategy"? Please explain and/or clarify.
>
I have added some text just before to explain that you might assume that
node failures with microloops were sufficiently that you might choose to
ignore the case.
>
>  1. Section 7. "Ideally this is provided by the remote LFA repair
>     target advertising this address in the IGP in use. . .out of scope
>     for this document and must be specified in an IGP specific RFC."
>     What's the importance of even mentioning this here? If the
>     extensions don't exist and there is a mechanism that does (using
>     the router ID, as mentioned), then why include something that
>     doesn't exist..? IMHO, if you want those extensions to be
>     available, then go define them and then point back to this draft
>     as a way to know where to establish the session.
>
>
This text was debated a lot and I would prefer to see it go as is to the
next review stage.
> Nits
>
>  1. Terminology. It would be nice to reorder this section so that
>     terms defined later are not used to define terms earlier. For
>     example, P-space is used in the definition on Extended P-space.
>  2. Introduction. s/many topologies[RFC6571]/many topologies [RFC6571]
>     (missing space) There are several other places where the space is
>     missing, which sort of screws up the references in the html
>     version..maybe something that the RFC Editor will pick up on..I hope.
>  3. Introduction. s/This technique describes in this document/The
>     technique described in this document
>  4. Repair Paths.
>     s/[I-D.ietf-rtgwg-lfa-manageability].]/[I-D.ietf-rtgwg-lfa-manageability].
> 5.
>     4.2 s/protected link S-E.,/protected link S-E,
>  6. 4.2 s/When released, tunneled packets/When released, packets (not
>     in the tunnel anymore)
>  7. 4.2.1 s/Finally we how to compute/Finally we show to compute
>  8. 4.2.1 s/is provide in/is provided in
>  9. 4.2.2 s/that member of the set/that the member of the set
> 10. 4.3 s/pseudo-code provides/pseudo-code provided
> 11. rfc3137 has been obsoleted by rfc6987
> 12. 7. s/for the emote LFA/for the remote LFA
>
1..12 done
>
>  1. Section 8.
>       * I always like results in an Appendix, but that's just
>         me..there may have been a discussion already about it (similar
>         to the pseudo-code).
>
It has been here for a while, can we see what we get in the next round
of reviews.
>
> 1.
>       * There are some places where some explanation could be useful:
>         what does the fact that the PQ session could not be
>         established mean? What is the significance of the percentiles?
>         Etc.. Just to have a more complete report.
>
It is the number of cases where neither LFA nor simple RLFA was adequate
- I have added some text.
>
> 1.
>       * The last few sentences of the section propose other solutions
>         in case the intersection between P and Q spaces is empty. I
>         think that is out of scope and should be taken out. As with
>         the suggestion of IGP extensions (section 7), if there is a
>         specific/complete solution then it should be specified.
>
I have trimmed this down to

In the small number of cases where there is no intersection between the
(extended)P-space and the Q-space, a number of solutions to providing a
suitable path between such disjoint regions in the network have been
discussed in the working group. For example an explicitly routed LSP
between P and Q might be set up using RSVP-TE or using Segment Routing
[I-D.filsfils-rtgwg-segment-routing]. Such extended repair methods are
outside the scope of this document.

> 1.
>  2. Section 10 ends with "The following section motivates..", but
>     there is no such section.
>
Text deleted

- Stewart
_______________________________________________
rtgwg mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/rtgwg

Reply via email to