Re: [mpls] Issues with your use of TLV Type and sub-types in your draft

[Loa Andersson <[e-mail]>]

Eric, Sami et.al. ,

I'd like to split this on two may be three steps.

1. Let us understand the situation, mostly have someone implemented and
    deployed the draft-ietf-mpls-tp-on-demand-cv

    a. if there is it is comparatively easy to defuse the situation
       we will need to agree with wg, IESG and IANA how to rearrange
       sub-TLV type assignments; this could be done quickly

    b. it here is we need to take other measures, we can't live with
       colliding sub-TLV types

2. Once this is sorted out we need to verify that the rest of the
    sub-TLV  type assignments are correctly done and that the rules
    for assignments are understood

3. We need to write some type of directive, that is includes e.g. when
    to request early assignments

/Loa


On 2010-10-28 09:14, Eric Gray wrote:
> George/Nitin/Kireeti,
> A comment on your draft, for information only, at this point.
> You have suggested a new TLV type (20) and 3 new TLV sub-types (1,
> 2 and 3) in your draft (draft-ietf-mpls-lsp-ping-enhanced-dsmap) that have
> some issues.
> First, we are rapidly discovering that "suggesting" actual values for these
> numbers may be a serious problem. There is a process for early allocation
> of codepoints managed by IANA. It is a simple process, requiring only that
> the draftco-authors request early allocation by the appropriate WG
> chair(s).
> The WGchair will take it from there.
> It is my opinion that - until assigned - these code points may only be
> referred to as "TBD."
> Second, the TLV type you have suggested is also suggested by another
> draft (draft-ietf-mpls-return-path-specified-lsp-ping), but for a
> different purpose.
> This is precisely the reason why "suggesting" values is a problem for IANA
> managed name spaces.
> Third, if you are defining a new TLV-Type from the name-space defined by
> RFC 4379, you can (and probably should) use the numbers in sequence
> (starting either with zero or one), in order to avoid confusing and/or
> fragmenting
> the number space unnecessarily.
> You have chosen to start with the value 1, but have not indicated a specific
> reason why zero should not be used. In fact, you have stated that the valid
> range is 0-65535. You may want to either use the values 0, 1 and 2, or state
> why zero is not used.
> Sinceyou are in fact, defining a new TLV type, you do not need to request
> IANA assignment of the sub-type numbers (since you have explicitly defined a
> new name space for the TLV type).You should simply make the assignment
> of these initial TLV sub-types part of your definition of the new
> sub-registry.
> The WG chairs are working with the IESG to resolve this and related issues.
> I believe it would be inappropriate to take any action on this
> information until
> you are informed by the WG chairs as to what action to take.
> --
> Eric

-- 


Loa Andersson                         email: [e-mail]
Sr Strategy and Standards Manager            [e-mail]
Ericsson Inc                          phone: [phone]
                                              [phone]