Minutes of the IEEE SSSWG Meeting
Held May 30-June 1, 2000
Held at the
Holiday Inn, Capitol Hill, Washington, D.C.
Evening Out at the
Nora's
Restaurant, DuPont Circle
Jack Cole, IEEE Sponsor and
SSSWG Chair
1244.1 Architecture/Data Model
Geoff
Peck (principal), Curtis Anderson, Joel Williams, Murali
Sathyanarayana editors
1244.2 Session Security, Authentication, and Initialization
Protocol
Jan Klier (principal), Bruce Haddon, Curtis
Anderson, Joel Williams
1244.3 Media Management Protocol
Murali Sathyanarayana (principal), Curtis
Anderson, Joel Williams
1244.4 Drive Management Protocol
Joel
Williams, editor
1244.5 Library Management Protocol
Joel Williams, editor
ATTENDANCE (7). Curtis Anderson Turbolinux canderson@turbolinux.com Jack Cole SELF jack.cole@ieee.org Merritt Jones MITRE merritt@mitre.org Paul Lockwood Legato plock@legato.com Geoff Peck Peck Labs, Inc. geoff@peck.com Alan Rollow Compaq alan@nabeth.cxo.dec.com Joel Williams SES joelw@ses-inc.com
Minutes from the Colorado Spring's meeting were approved after amendments were made in accordance with Geoff's suggestions.
In a general group discussion of increasing SSSWG participation, Geoff suggested that we set up a mini-symposium on just 1244 and promote it as an industry conference.
Discussions of the next media management system (MMS) standards and Mover (not part of MMS) lead to these points:
* 1244.6 - One suggestions is to use XML while another is to parse the output of a set of SHOW commands and create the corresponding stuff locally. Joel will do a strawman using SHOW commands. Curtis will do an example of XML usage.
* 1244.7 - Deferred for now. Seems to be decreasing in importance relative to the other work items.
* 1244.8 Paul will take capi.h and update all the names and add manpages for the functions and post the result as a strawman.
* 1244.9 Geoff will expand on his strawman where UMA puts attributes on each volume in a format such as "hostname:user:password:priviledge" where any portion can be wildcarded.
* 1244.10 - Paul proposed a structure of one command per object type in the data model with list, create, delete, and modify options. He will email his strawman. Geoff proposed a CVS-like interface of one command with subcommands and options. Alan will email a CVS-like strawman to the group.
* 1244.11 - Curtis presented some half-formed ideas on how it could be done and the group debated the pro's and con's. No conclusions or action items. The proposal is to separate the control flow from the data flow so that control operations are in one channel while pure data is in the other channel. Separate channels was seen as a double-edged sword in that there was a risk of getting out of synchronization but it had the ability to route the data flow through other media (eg: fibrechannel) transparently to the app. Given that simpler is better, Geoff suggested that 1563 start from the existing RMT protocol. There is no need for striping support (ala HPSS) and the API must use a standard interface.
The group listed (most) of the ideas/features that we have talked about before that were consciously deferred to version 2 of the MMS standards:
* distributed atomic mount operations
* finer grained, capability based, security model
* callback event notification instead of polling
* disconnecting/reconnecting from sessions
* adding more detail on manual operations
* automatic inter-library transfers of cartridges
* LM-based drive selection
These will be considered at each revision or amendment of the standards, and may appear in later reballoted versions of the standards.
Ken Oettle made a presentation to the group on JASA's (Joint Airborne Signal Intelligence Architecture) interests in storage standards. See www.jswg.org for more details.
1563.1, 1563.2, 1563.3 - Curtis presented the current state and proposal for each of the 3 PARs. There was minor discussion on .1 and .2 and lots of debate on .3. Geoff says that there are much better Data Description Languages than XML and that 1563 should do research and compare. [Note: On the Friday after the SSSWG meeting, Joel, Curtis, and Ben Kobler meet at AIIM headquarters and discussed the 1563 standards. Joel and Ben Kobler explored the question of how much overlap there is between the AIIM MS66 standard for accessing HSM catalog information and 1563.3. The conclusion was that there was significant overlap, but that 1563.3 was more powerful (if it could be made to work). For example MS66 can't describe the OpenTape format that Legato NetWorker uses.]
In a discussion of meeting frequency, the
membership seemed to favor sticking with the six-per-year
frequency over a reduced schedule of four-per-year. Jack again
emphasized the need to get work done outside of meetings, and
that work of this group is not focused on meetings. And the
cycling of meeting locations between Silicon Valley, Colorado,
and Maryland-D.C. will continue at least through 2000.
Next Meeting. July 11-13, 2000, Palo Alto, CA, Hosted by Legato
Future Meetings.
September 19-21, 2000 (Boulder or Louisville, Hosting Sought)
November 14-16, 2000, Aberdeen, MD (HEAT Center)