SSSWG
March 1998 Minutes
Preliminary Meeting Minutes for March 1998
IEEE Storage Systems Standards Working Group (P1244)
Held at the National Center for Atmospheric Research (NCAR),
Damon Room, Boulder, Colorado March 10-12, 1998
IEEE Sponsor Merritt Jones
SSSWG Chair Jack Cole
MSSRM Tech Ed Sam Coleman
MMS Editor Geoff Peck
Co-Treasurer Joel Williams
Attendance (12).
----------------------------------------------------------------------
Curtis Anderson SGI curtis@sgi.com
Jack Cole ARL jack.cole@ieee.org
Sam Coleman LLNL scoleman@llnl.gov
Brian Dodd EASTMAN brian.dodd@eastmansoftware.com
Bruce Haddon REDCAPE bkh@spot.colorado.edu
Dixon Hutchinson EXABYTE hutch@exabyte.com
Paul Lockwood LEGATO plock@legato.com
John Merrill NCAR jhm@ucar.edu
Kati Nelson LEGATO knelson@legato.com
Geoff Peck QUANTUM geoff@peck.com
David Richey QUANTUM richey@richey.com
Eric Stouffer IBM stouffer@us.ibm.com
Thanks.
------------------------------------------------------------------------
We all greatly appreciated the spectacular meeting room, tour of NCAR
storage facility, and other benefits provide to the SSSWG by NCAR.
Special thanks to John Merrill for arranging this support.
We happily acknowledge NCAR support to SSSWG at past meetings, beginning
with the first SSSWG meeting.
Administrative.
------------------------------------------------------------------------
The meeting began at 8:30 am on Tuesday with the usual business of
attendance and registration. Minutes from the Las Vegas, NV meeting were
approved along with the agenda for this meeting. The meeting ended
at noon on Thursday.
Registration Fee. It was decided to raise the fee to $150.00. The
purpose of the fee is to cover costs, not to make a profit. But the
present fee is years old, and we are far from covering costs at $90.
The new fee is set based on annual average costs in hosted meetings and
meetings held at hotels, commercial conference facilities. It will
probably not entirely cover SSSWG costs, but it will slow the rate
at which SSSWG is getting behind financially to the point that further
increases can be delayed for years.
Identity of Voting Members. Many requested that the voting members be
identified on the SSSWG web pages, in part so that they may serve as
voices for individuals unable to attend meetings. It will be done.
Rename PAR. The P1244.2 PAR was submitted as "Standard for Media
Management System (MMS) Session Security, Authentication, and
Initialization Protocol (SSAIP)", but it is desired to change the
acronym for this component simply to "SIP". That will be accomplished
at a convenient point in time.
PR/Tutorials/November Meeting. Public relations, press releases were
discussed. Jack is working on the latter. Concern was expressed that
SSSWG attendance might increasingly require too much time for tutorial
to newcomers. In this regard, the Group entertained the possibility of
holding a day in advance of each meeting for such tutorials, but if and
only if required by individuals registering their need in advance. When
the Group started to consider dates for future meetings, and the
conflict in November between the SSSWG meeting and Supercomputing 98
dates became apparent, Geoff suggested that an MMS tutorial could be
given at SC98. He volunteered to give the tutorial, and Jack said he
would arrange with the Supercomputing organization for the tutorial to
be held. The thought is that people attending the November SSSWG
meeting who desire a tutorial will be sent to the SC98 tutorial. In
addition, since SC98 and SSSWG are holding meetings in Florida in
adjacent weeks, the SSSWG may begin its meeting on Sunday following
SC98.
New Meeting Announcements. As a direct result of the concern over the
need for tutorials, and to allow for better meeting planning, the
announcement of the May and subsequent meetings will be made differently
than for past meetings. Pre-registration might be required and time in
tutorial not permitted during the regular meeting because of the tight
schedule that SSSWG has in getting standards to ballot in 1998.
MEETING.
------------------------------------------------------------------------
Geoff gave SSSWG the talk he presented at SNIA and NSIC/NASD on MMS, and
talked about object storage, object oriented disks, and the NASD WG
viewpoint of net attached {storage|secure} {devices|disks}. Slides for
these and other relevant presentations can be viewed at the following
URLs:
http://www.nsic.org/nasd/1998-mar/nasd-ns1.pdf
http://www.nsic.org/nasd/1998-mar/nasd-nsi.pdf
http://www.peck.com/~geoff/peck-snia-980305.pdf
http://www.snia.org/398pres.htm
The ideas presented were so interesting to the group that there were
many questions in the morning, and regaining focus on the WG short term
goals in the afternoon was very difficult. The idea of making disks
carry a great deal of the work normally done by servers lead the group
into a discussion about security and making disks cache server
information in parallel (striping). Geoff stated that this has been
considered, and mentioned the Parallel Data Labs (CMU) Cheops Clerk.
Examine the information at the URL for Scalable I/O at the PDL:
http://www.pdl.cs.cmu.edu/NEW/SIO/SIO.html
Other topics such as spindle synch and read ahead, pre-fetching were
discussed as well. Interestingly, NASD has no real 'NASD', but uses
disk- Alpha Workstation pairs as NASDs, for the reason that no disks
with appropriate microcode exist (yet).
On the topic of storage objects, software downloads to disks was
discussed along the lines of protection of the disk software, the
possible tyranny of disk drive manufacturers (discounted because
of marketplace competition), and the possibility of integrating
applications (by download) such as tar to devices such as tape drives.
The idea of apps residing on NASDs raised the eternal question Œof who
the principal is in such interactions.
Discussion of "Data Mover" was re-opened in the context of NASDs. There
is consensus that a naive MOVER would be subject to snooping/spoofing;
that MOVER is really a collection of methods; and that MOVER
functionality must exist. But there is a divergence of views on where
the functionality must/will reside. TBD.
Brian volunteered to try to advance the definition of MOVER, and Jack
said he would send Brian background information which Brian can use
as he wishes. Or start fresh.
MMS Requirements were discussed very briefly with no resulting changes
to the document. However, Requirements will be subject to continuing
adjustment as needed with the purpose of the document remaining to serve
the WG as reference for development of the MMS.
The MMS Architecture Document was discussed. A very well written
document to begin, most of the effort on this document will now focus on
making it more readable to outside audiences. To accomplish this, Geoff
began adding sections to the beginning and revising, adding drawings as
suggested by the group. A Group edit of the document was performed by
projection of the live document to screen. This unfortunately required
closing blinds to reduce outside light on the screen, and eliminating
the view of the Flatirons :-( A curious insertion of a reference to
farms was made to the document, and the SSSWG anxiously awaits public
comment.
An issue which arose in discussion of the MMS architecture was whether
fixed media was precluded. It was decided to make slight changes to the
Architecture document to say nothing precluding fixed media.
Repeatedly discussion arose about MMS versus an Object Storage Model
and about MOVER as standalone functionality, functionality sprinkled
throughout MMS, or functionality completly overcome, subsumed within an
Object Storage Model.
Because of the attractiveness of a lot of ideas new to many in the
group, focus was difficult to maintain. But reminding ourselves of the
clear fact that SSSWG is targeting five standards from MMS for IEEE
balloting in 1998 had a very sobering effect.
Curtis provided a synopsis of updates, changes, and status of SGI's
OpenVault, now at version 1.1.
EVENING OUT.
------------------------------------------------------------------------
Normally the evening out is held on Wednesday, but it fit members'
schedules better to hold the dinner on Tuesday. The WG dined at the
Briarwood Inn in Golden, Colorado, and Jenny, Dixon's wife and Toula,
John's wife were able to join us. Photos of the group were taken and
will be placed on the SSSWG web site. We had a great time!
GROUP PICTURE.
------------------------------------------------------------------------
Photos of the group at NCAR were taken, and will appear at the SSSWG web
site before the next meeting. Likewise, a group photo from the Las Vegas
meeting in January and for future meetings will appear at the site.
PROCESS TO DEVELOP DOCUMENTS FOR THE MMS COMPONENT STANDARDS.
------------------------------------------------------------------------
Only the Requirements doc has been opened to public review, so far. But
it is expected that all documents will be available to SSSWG members for
review by March 27th. And by May all documents will available for
public comment.
A table will be displayed at the SSSWG web site for SSSWG members to view
showing the status and contact for each document.
As documents become available, minor revisions should be sent as
annotated hardcopies by fax to the appropriate person.
Extensive comment or revision should be sent in Word 7 or
Office97 Word documents by email to the appropriate editor, herder
of the document.
An anonymous write-only ftp site may be set up for this purpose.
The documents to be developed immediately are ARCH/DB, SIP, MMP, LMP, DMP.
P1244.1 Standard for Media Management System (MMS) Architecture,
Including Database
P1244.2 Standard for Media Management System (MMS)
Session Security, Authentication, and Initialization Protocol
(SSAIP)
P1244.3 Standard for Media Management System (MMS)
Media Management Protocol (MMP)
P1244.4 Standard for Media Management System (MMS)
Drive Management Protocol (DMP)
P1244.5 Standard for Media Management System (MMS)
Library Management Protocol (LMP)
Timeline (for development of the documents listed above).
----------------------------------------------------------------------
March 27, 1998 - Initial Versions to SSSWG Members, Editors,
Herders Appointed
May 1, 1998 - First Public Versions of Each Document Available,
Input from Public Accepted
Submit MMS to IEEE Ballot Process by September 98 Meeting
Next Meeting.
----------------------------------------------------------------------
May 12-14, 1998, QUANTUM, Milpitas, CA (Geoff Peck)
Future Meetings.
----------------------------------------------------------------------
July 14-16, 1998 Redmond, WA (Jonathon Otis/ADIC)
Sept 15-17, 1998 Maryland (Jack Cole/ARL)
Nov 9, 1998 MMS Tutorial at Supercomputing '98, Orlando, FL
Nov 17-19, 1998 *OR* Nov 15-17, 1998, Key West, FL
SSSWG Home