Re: [sfc] NSH MD-1 documents

[Tal Mizrahi <[e-mail]>]

Hi,



We have posted an updated version of
draft-mymb-sfc-nsh-allocation-timestamp.

https://tools.ietf.org/html/draft-mymb-sfc-nsh-allocation-timestamp-02



The use cases for this allocation are described in Section 4 (
https://tools.ietf.org/html/draft-mymb-sfc-nsh-allocation-timestamp-02#section-4
).




Two notes:

1.       It is a general property of MD Type 0x1 that there is an
underlying assumption that either the control plane  or the management
plane take care to determine (and configure) which specific MD 0x1
allocation is used. Presumably, in a given network only a single MD Type
0x1 allocation will be used, as there is no way for SFs to distinguish
between different MD Type 0x1 allocations.

2.       Having said that, we still believe that there may be a good case
for having a few different allocations, which may fit different operator
requirements. Some operators may require a high-resolution timestamp and a
sequence number, and will choose to use
draft-mymb-sfc-nsh-allocation-timestamp-02. Other operators may choose an
allocation such as draft-guichard-sfc-nsh-dc-allocation-07, that includes
other metadata fields and an optional lower-resolution timestamp. We
believe that other allocations may also benefit from including a (few bits
of the) timestamp field.



Comments will be welcome.

Cheers,

Tal Mizrahi.

On Wed, Aug 2, 2017 at 6:37 PM, Joel M. Halpern <[e-mail]> wrote:

> There have been a number of documents brought forward describing layouts
> of the 16 byte MD-1 metadata for specific situations.  The working group
> has discussed publishing some set of these as informational RFCs.  We would
> like to progress this work.  Without either overloading the working group
> or serving as a rubber stamp for proposals.
>
> As such, would authors of MD-1 proposals that they believe should be
> worked on by the WG please do two things:
> 1) Refresh your documents, so we know they reflect your current view.
> 2) Provide an explanation to the WG of why this proposal has sufficiently
> broad applicability to warrant WG effort.  Not just "Data Centers are
> important".  But something that gives the WG members more context as to why
> the document addresses a broad range of cases.  We are not looking for
> detailed use cases, but a summary in email of the key points would help
> folks decide what to read.
>
> Thank you,
> Joel and Jim
>
> _______________________________________________
> sfc mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/sfc
>