Re: [storm] Publication requested for RDDP registries draft

[<[e-mail]>]

Hemal,

> Thanks for the reply! Your points #1 and #2 suggest that we need to revise RDMA protocol extensions draft with IANA registry section. Please confirm.

That is correct.

> After the IANA registries for RDDP draft goes to RFC, what is the estimated time for the actual creation of IANA registries for RDDP (that is accessible from IANA website)?

The registries will appear almost immediately after RFC publication, but it is not necessary to wait until then to revise the RDMA Protocol Extensions draft.  See the IANA Considerations section of the MPA Peer Connect for an example of language that adds entries to a registry that will be created by another draft.

Thanks,
--David

From: Hemal Shah [mailto:[e-mail]]
Sent: Friday, November 18, 2011 7:51 PM
To: Black, David; [e-mail]
Subject: RE: Publication requested for RDDP registries draft

David,

Thanks for the reply! Your points #1 and #2 suggest that we need to revise RDMA protocol extensions draft with IANA registry section. Please confirm.

After the IANA registries for RDDP draft goes to RFC, what is the estimated time for the actual creation of IANA registries for RDDP (that is accessible from IANA website)?

Hemal

From: [e-mail] [mailto:[e-mail]]
Sent: Thursday, November 17, 2011 2:25 PM
To: Hemal Shah; [e-mail]
Subject: RE: Publication requested for RDDP registries draft

Hemal,

No change is needed to the RDDP registries draft, as all three of your items belong elsewhere:

1: The new RDMAP message opcodes will be added to the IANA registry by the RDMAP protocol
extensions draft when it is published as an RFC.
2: The RDMAP protocol extensions draft is the appropriate place to create a new
Atomic operations code IANA registry for RDMAP.
3: In the future, only the IANA registry is updated.  Any standards track RFC can add
      new registry entries and even create new registries.  The IANA registries are the
      definitive reference; the RDDP registries draft is needed only because the registries
      weren't created when the RDDP protocols were originally standardized.

In fact the registries in the RDDP registries draft are "incomplete" in another way, as they do not contain any entries from the MPA Peer Connect draft.  This is deliberate: after the RDDP registries RFC creates the registries, the MPA Peer Connect RFC will add entries to two of them.  In the future the RDMA protocol extensions RFC will add entries to another one of these registries, and it looks like you want to add text to create an atomic operation codes registry.

Items 1 and 2 are premature, as there is no final version of the RDMA protocol extensions draft.  IANA Considerations text will need to be added to the RDMA protocol extensions draft to address items 1 and 2.

Thanks,
--David
----------------------------------------------------
David L. Black, Distinguished Engineer
EMC Corporation, 176 South St., Hopkinton, MA  01748
[phone]             FAX: [phone]
[e-mail]        Mobile: [phone]
----------------------------------------------------

From: Hemal Shah [mailto:[e-mail]]
Sent: Thursday, November 17, 2011 12:49 PM
To: Black, David; [e-mail]
Subject: RE: Publication requested for RDDP registries draft

David,

Sorry to respond late to this thread! I have several comments on the IANA registries for RDDP draft before it goes to the publication.


 1.  The current draft does not include the RDMA Message Opcodes defined in the RDMA protocol extensions draft (Figure 3 of http://www.ietf.org/id/draft-ietf-storm-rdmap-ext-01.txt). I would like to see them included in the draft below.
 2.  The RDMA protocol extensions draft defines several Atomic operations that are identified by Atomic operations codes (Figure 5 of http://www.ietf.org/id/draft-ietf-storm-rdmap-ext-01.txt). I believe it will be good to add them to the IANA registries for RDDP draft.
 3.  For any new opcodes in the future, do we need to update both relevant RFCs and registry specs? Or, registry spec only? If the opcode does not change any semantics in the RFC, then it would make sense to only update the registry spec.

Regards,

Hemal

[... remainder snipped ...]