Re: [Gen-art] Gen-ART Last Call review of draft-ietf-pcp-port-set-10

[Meral Shirazipour <[e-mail]>]

Hi,
  Thank you Mohamed, I am ok with the clarifications and changes.

Best,
Meral

> -----Original Message-----
> From: [e-mail]
> [mailto:[e-mail]]
> Sent: Tuesday, October 13, 2015 11:32 PM
> To: Meral Shirazipour
> Cc: [e-mail]; [e-mail]
> Subject: RE: Gen-ART Last Call review of draft-ietf-pcp-port-set-10
> 
> Hi Miral,
> 
> Please see inline.
> 
> Cheers,
> Med
> 
> > -----Message d'origine-----
> > De : Meral Shirazipour [mailto:[e-mail]]
> > Envoyé : mardi 13 octobre 2015 18:31
> > À : BOUCADAIR Mohamed IMT/OLN
> > Cc : [e-mail]; [e-mail]
> > Objet : RE: Gen-ART Last Call review of draft-ietf-pcp-port-set-10
> >
> > Hi,
> >   Many thanks for the clarifications. Please see inline.
> >
> > Best,
> > Meral
> >
> > > -----Original Message-----
> > > From: [e-mail]
> > > [mailto:[e-mail]]
> > > Sent: Tuesday, October 13, 2015 4:42 AM
> > > To: Meral Shirazipour; [e-mail];
> > > gen- [e-mail]
> > > Subject: RE: Gen-ART Last Call review of draft-ietf-pcp-port-set-10
> > >
> > > Dear Meral,
> > >
> > > Many thanks for the review.
> > >
> > > Please see inline.
> > >
> > > Cheers,
> > > Med
> > >
> > > > -----Message d'origine-----
> > > > De : Meral Shirazipour [mailto:[e-mail]]
> > > > Envoyé : lundi 12 octobre 2015 07:49 À :
> > > > [e-mail]; [e-mail] Objet
> > > > : Gen-ART Last Call review of draft-ietf-pcp-port-set-10
> > > >
> > > > 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 resolve these comments along with any other Last Call
> > > > comments you may receive.
> > > >
> > > > Document: draft-ietf-pcp-port-set-10
> > > > Reviewer: Meral Shirazipour
> > > > Review Date: 2015-10-10
> > > > IETF LC End Date:  2015-10-14
> > > > IESG Telechat date: NA
> > > >
> > > >
> > > > Summary:
> > > > This draft is ready to be published as Standards Track RFC but I
> > > > have some comments.
> > > >
> > > >
> > > > Major issues:
> > > > N/A
> > > >
> > > > Minor issues:
> > > > -[Page 7], Section 4.1, "If the PCP Client does not know the exact
> > > > number of ports its requires, it MAY then set the Port Set Size to
> > > > 0xffff, indicating that it is willing to accept as many ports as
> > > > the PCP server can offer."
> > > > Question/clarification to add: Mention if there a mechanism where
> > > > the server will know which of the mapped ports are going to be
> > > > used by the client? and which mappings can be discarded/reused in
> > > > a subsequent request.
> > > >
> > >
> > > [Med] I'm not sure to get your point, especially the link with the
> > sentence
> > > you quoted. But fwiw policies about granted port ranges (size),
> > > ports to
> > be
> > > excluded from assignment, etc. are implementation-specific. These
> > > are similar to the behavior of the PCP server assigning single port
> > > numbers (RFC6887). If the question is about renewal and/or port
> > > overlap, the
> > behavior
> > > is called out in Section 4.4.
> > >
> >
> > [MSh] Sorry I was a bit unclear here. I was wondering what happens if
> > the client asks with 0xffff, receives e.g. 100 ports but only uses 20 of them.
> > What happens to the rest? Would they be unused until the next renewal?
> > If so efficiency is not affected?
> > [please also see the thread reply by Simon]
> 
> [Med] Thank you for clarifying. The size of port ranges that are assigned to
> PCP clients is deployment-specific. Operators will need to tune the maximum
> size of the port sets to be assigned taking into account various inputs such as:
> optimize the use of the shared addresses, reduce the amount of pcp
> messages, etc. Of course, efficiency will depend on the size of the assigned
> port set and the actual usage from this set. This is exactly the same issue with
> setting a port quota.
> 
> FWIW, below is provided a sample YANG excerpt to configure the port set
> feature in a PCP server.
> 
>    grouping port-set-option {
>        description
>                     "PORT_SET option.";
> 
>        leaf port-set-enable {
>           type boolean;
>           description
>              "Enable/disable PORT_SET option.";
>        }
> 
>        leaf default-port-set-size {
>           type uint16;
>           description
>              "Indicates the default size of a port set.";
>        }
> 
>        leaf maximum-port-set-size {
>           type uint16;
>           description
>              "Indicates the maximum size of a port set.";
>        }
>    }
> 
> It is up to each operator to set those parameters. Traffic analyses are likely to
> help operators to set the appropriate values.
> 
> The port-set draft does not need to deal with these aspects as those are
> deployment-specific.
> 
> >
> > > >
> > > > Nits/editorial comments:
> > > > -[Page 6], "In particular, the PREFER_FAILURE option MUST NOT be
> > > > present in a request that contains a PORT_SET option.".
> > > > Suggestion: Please add a sentence after this one suggesting why
> > > > PREFER_FAILURE option MUST NOT be used. It was not clear to me
> > > > until I read the rest of the draft...although I am still not sure
> > > > why this behavior is to be a MUST NOT.
> > >
> > > [Med] PREFER_FAILURE was specifically designed for the interworking
> > > with UPnP IGD:1 (RF6970). The rationale why it should not be used by
> > > other
> > PCP
> > > clients (than the iwf) is discussed in Section 13.2 of RFC6887. The
> > language in
> > > this draft is stronger than RFC6887, though. The decision about the
> > language
> > > to use was made in an interim meeting (see the minutes at:
> > >
> http://www.ietf.org/proceedings/interim/2014/08/26/pcp/minutes/minut
> > > es -interim-2014-pcp-1). We can add the following sentence.
> > >
> > > NEW:
> > > As a reminder PREFER_FAILURE was specifically designed for the
> > > Universal Plug and Play (UPnP) Internet Gateway Device - Port
> > > Control Protocol Interworking Function (IGD-PCP IWF) [RF6970]. The
> > > reasons for not recommending the use of PREFER_FAILURE are discussed
> > > in Section 13.2 of [RFC6887].
> > >
> >
> > [MSh] That looks great, thank you.
> 
> [Med] This will be added to the next revision of the draft.
> 
> >
> > > >
> > > > -[Page 8], Section 4.3, "There is intentionally no port set
> > > > capability discovery mechanism.".
> > > > What is the intention?
> > >
> > > [Med] This sentence is there to explicitly call out that this
> > specification does
> > > not define a mean to discover whether the PCP server support the
> > > port
> > set
> > > capability. As a generic comment, the working group felt it is early
> > > to
> > define a
> > > mechanism to retrieve the capabilities of the PCP server (and the
> > > PCP- controlled device): see the minutes available here:
> > > http://www.ietf.org/proceedings/85/minutes/minutes-85-pcp (search
> > > for draft-boucadair-pcp-capability).
> > >
> > >  I could not find anything on the list discussion.
> > > > It would be good to clarify this to make this section puroposeful.
> > > >
> > >
> > > [Med] This section was added to address the comment recorded in the
> > > minutes: http://www.ietf.org/proceedings/88/minutes/minutes-88-pcp.
> > > See the changes here:
> > > https://www.ietf.org/rfcdiff?url1=draft-ietf-pcp-port-
> > set-
> > > 04&url2=draft-ietf-pcp-port-set-05
> > >
> >
> > [MSh] Thank you for the pointers, so if I understand correctly the
> > reason is "because they cause unnecessary failures" ?
> >
> 
> [Med] Yes, that was the main issue raised at that time. After re-reading the
> text, I think we can delete "There is intentionally no port set capability
> discovery mechanism.", the rest of the paragraph is clear enough.
> 
> >
> > > > -[Page 16] ,  Ref. [RFC7596] should be revised-it still refers to
> > > > the draft
> > >
> > > [Med] Thank you for catching this. This will be fixed.