July
2000 Minutes
IEEE
Storage System Standards
Working Group
(P1244)
Minutes
Meeting Held July 11-13, 2000
At Legato Corporation, Palo Alto, California
Evening Out at Chantilly's Restaurant, Atherton
Curtis Anderson, SSSWG Chair, presiding
Jack Cole, IEEE Sponsor
Attendance (5)
Curtis Anderson, Turbolinux, canderson@turbolinux.com
Bruce Haddon, Sun, bkh@spot.colorado.edu
Paul Lockwood, Legato, plock@legato.com
Alan Rollow, Compaq, alan@nabeth.cxo.dec.com
Joel Williams, SES, joelw@ses-inc.com
* Minutes from the Washington, DC, meeting were approved without change.
* The next meeting schedule was confirmed for 9/12-9/14. Bruce will try
to convince Sun to host the meeting at their Boulder, CO. facility. Other
options include a hotel in the Boulder area, or possibly in Colorado
Springs. We will finalize this via email.
* We worked on the responses to the outstanding ballot comments, deciding
on recommended wordings to be sent to the 1244 Ballot Review Committee.
* 1244.2 - After long discussion of the comments received during the
ballot, we saw no compelling advantage to Message Integrity Coding (MIC)
and several advantages to SSL. We decided not to propose any changes to
the balloted 1244.2. Bruce took copies of the proposed responses plus
notes from the day's discussion and will email a final proposed response
for the 1244.2 comments in about a week. Those proposed responses will
be reviewed by the 1244 Ballot Review Committee for a final decision.
* 1244.6 - We continued discussion of the options presented at the last
meeting. We decided that it should support export/import of any MMS
information, not just cartridges, and that we would pursue a very simple
XML encoding of that data. Under this proposal an administrative
application would collect the information it desired to export from the
MMS using the normal MMP defined command set, then it would encode that
data into an XML description. Joel agreed to generate a strawman of
this approach and email it for group review.
* 1244.10 - Alan presented his initial proposal and we debated. Alan
will prepare another version of his proposal incorporating the feedback
from this meeting. We decided that the user interaction style of 1244.9
and 1244.10 must be very similar. Use of 1244.9 to mount cartridge will
be common for both operators and system administrators as well as users.
* 1244.11 - We had a long discussion of how the output of the IETF's
nascent iSCSI working group will relate to the Mover. We took note of
NDMP being similar to iSCSI in some respects. We decided that we needed
a clearer statement of the requirements for the Mover so that we could
evaluate whether there was sufficient need for us to develop a protocol
that would not be compatible with those other efforts. We started work
on that document.
* Futures List, "Callbacks" - We worked out a proposal on how to
support
an event model, including registration, notification, and cancellation.
* Futures List, "LM-Based Drive Selection" - We decided that the
current
standard does not preclude this, so we may be done. It needs a bit more
thought to verify that it can be supported.
* Futures List, "Automatic Interlibrary Transfer" - We discussed it at
length, but came to no conclusions. Modeling all possible offsite
storage policies (even at a very non-detailed level) is daunting. Some
level of support may be possible with the current standards if an
administrative application were to define its own attributes on the
cartridges under management.
* Futures List, "Central Configuration Management" - This is the
ability
of a person at a central console to administer the configuration of an
individual LM or DM. An example is the ability to change the pathname
that the LM uses to access the library. This capability is not directly
available via the LMP. Legato has done some research and development
work in this area and Paul will check to see if they are willing to
share that with the group.
* Futures List, "Session Disconnect and Reconnect" - We developed a
detailed proposal on how to do this with minimal impact on the existing
standards.
* Futures List, "Fine Grained Security" - We discussed a
capability-based
security model, including talking about exactly what that means, how it
might be implemented, and what the demand for such a complex approach is.
We decided that given the complexity, we would wait for more information
on the amount and type of demand there is in the industry for this before
proceeding.
Next Meeting. September 12-14, 2000, Boulder, Colorado, USA
Future Meetings.
November 14-16, 2000, Aberdeen, MD (HEAT Center)