[nfsv4] draft-haynes-nfsv4-layout-types-01

[David Noveck <[e-mail]>]

*From draft-haynes-nfsv4-layout-types-01:*

> Some implementers have interpreted the text in Sections 12 <http://tools.ietf.org/html/draft-haynes-nfsv4-layout-types-01#section-12> ("Parallel
> NFS (pNFS)") and 13 ("NFSv4.1 as a Storage Protocol in pNFS: the File
> Layout Type") of [RFC5661 <http://tools.ietf.org/html/rfc5661>] as both being strictly for the File Layout
> Type.

I don't see how one can read section 12 and think that it applies only
to the file layout.

After all 12.2.5 and 12.2.7 do explicitly mention block and object
layouts.  It may be that people are interpreting Section 12 as
referring only to existing layout types and not to those not yet
thought of when RFC 5661 was proposed.

I suspect that the root of the problem here  is that when RFC5661 was
published, people were not inclined to specify the sort of
requirements for layout type semantics that Tom puts forward in his
draft since it was generally agreed that block and object layout types
would be finished soon and would have the appropriate semantics.

I think Tom has identified a problem that we must address if we are to
be able to consider new layout types.  We have to have some
well-understood set of semantic requirements in order to make an
informed decision as to whether the proposed layout type meets the
requirements.

I think the requirements that Tom has articulated make sense.  If
people have issues with them, they need to comment now to start the
process of getting to a consensus on this set of issues.

I don't see how we can turn any proposal for a new mapping type into a
working group document until there is working group agreement on the
appropriate semantic requirements