Re: [core] Genart last call review of draft-ietf-core-sid-15

[Andy Bierman <[e-mail]>]

On Tue, Mar 16, 2021 at 10:34 AM Linda Dunbar via Datatracker <
[e-mail]> wrote:

> Reviewer: Linda Dunbar
> Review result: Not Ready
>
> I am the assigned Gen-ART reviewer for this draft. The General Area
> Review Team (Gen-ART) reviews all IETF documents being processed
> by the IESG for the IETF Chair.  Please treat these comments just
> like any other last call comments.
>
> For more information, please see the FAQ at
>
> <https://trac.ietf.org/trac/gen/wiki/GenArtfaq>.
>
> Document: draft-ietf-core-sid-15
> Reviewer: Linda Dunbar
> Review Date: 2021-03-16
> IETF LC End Date: 2021-03-17
> IESG Telechat date: Not scheduled for a telechat
>
> Summary:
> This document introduces the globally unique 63-bits integers to identify
> YANG
> items which include data nodes, RPCs, their associated inputs/outputs,
> notifications, YANG modules, submodules, and features.
>
> Major issues:
> - One of the benefits of YANG is its explicit naming and
> human-understandable
> notations. It will be a nightmare if all the YANG items are represented by
> integers. YANG items being represented by integers will be worse than the
> TYPE
> values in the TLVs. At least, the TLV types are in the context of the
> protocols
> and their messages. - It will be a tremendous amount of work to map all
> YANG
> items to globally unique integers. - YANG has been widely deployed without
> the
> numbering system. Which environment will need the integer representations?
> Who
> will validate the numbers used? How to validate the YANG modules
> represented by
> those "integers"?
>
>

My initial reaction to this work was that it would never fly in the wild.
But the CORE WG has developed tools and solved many of the problems with
YANG SID.

There is a great deal of trust required in the tooling and procedures used
to manage the mapping of
YANG identifiers to globally unique integers.  These numbers are not
intended for
human usage (much like OID strings in SNMP).  If the client and server do
not agree
on the YANG-to-SID mapping, then the protocols will not work (same as OIDs
in SNMP).
I think the administration plan makes sense, but it is still unproven,
especially the
delegated subrange name assignments.

The mapping files are independent of the YANG modules so there is no need
to alter
any YANG module to use YANG SID.  There is a lot of work creating mapping
files instead. ;-)
The mappings are not used in YANG modules at all. They are only useful to
generate an
efficient binary message encoding for YANG data.  (Only. The delta-SID
encoding is extremely efficient
and really needed right now.)


Andy


- SID can be mistaken as Segment Identifiers (SID). Suggest using a
> different
> name.
>
> Minor issues:
>
> Nits/editorial comments:
>
> Best Regards,
> Linda Dunbar
>
>
> _______________________________________________
> core mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/core
>