Accredited Standards Committee X3, Information Processing Systems Doc. No.: X3T9.2/92-230 R0 Date: December 9, 1992 Project: Ref. Doc.: Reply to: L. Lamers To: Membership of X3T9.2 From: Lawrence J. Lamers, Secretary X3T9.2 George Penokie Chair RAID Study Group Subject: Minutes of RAID December 7, '92 Study Group Meeting Agenda 1. Opening Remarks 2. Attendance and Membership 3. Approval of Agenda 4. RAID Models 5. Data Attributes 6. Device Identification 7. Physical Addressibility 8. Reconstruction Operations 9. Determining Devices State 10. INQUIRY data 11. LOG SENSE data 12. REQUEST SENSE data 13. Dynamic addition or deletion of capacity (lun) 14. Replacement drives 15. Meeting Schedule 16. Adjournment Results of Meeting 1. Opening Remarks George Penokie the Chair, called the meeting to order at 4:00 pm, Monday December 7, 1992. He thanked Rich Bailly of Vitro Corp. for hosting the meeting. As is customary, the people attending introduced themselves. A copy of the attendance list was circulated for attendance and corrections. It was stated that the meeting had been authorized by X3T9.2 and would be conducted under the X3 rules. Ad hoc meetings take no final actions, but prepare recommendations for approval by the X3T9.2 task group. The voting rules for the meeting are those of the parent committee, X3T9.2. These rules are: one vote per company; and any participating company member may vote. The minutes of this meeting will be posted to the SCSI BBS and the SCSI Reflector and will be included in the next committee mailing. George stated that this is the second meeting of the RAID study group. The purpose of the group is to deal with interface issues related to using RAIDs. The study group will assess the issues and then formulate a strategy for dealing with them. 2. Attendance and Membership Attendance at working group meetings does not count toward minimum attendance requirements for X3T9.2 membership. Working group meetings are open to any person or company to attend and to express their opinion on the subjects being discussed. The following people attended the meeting: RAID Study Group Meeting Attenders Name S Organization Electronic Mail Address ---------------------- - ------------------------- ------------------------- Mr. Norm Harris P Adaptec, Inc. nharris@adaptec.com Mr. Scott Smyers P Apple Computer smyers.s@applelink.apple. com Mr. Edward Haske P CMD Technology Mr. Stephen R. Cornaby P Conner Peripherals Mr. Soon Ahm Kang P Control Data Systems, soon.a.kang@cdc.com Inc. Mr. Bob Solomon O Data General Corp. Mr. Charles Monia P Digital Equipment Corp. monia@starch.enet.dec.com Dr. William Ham A Digital Equipment Corp. ham@subsys.enet.dec.com Mr. William Dallas A Digital Equipment Corp. dallas@zk3.dec.com Mr. Ralph Weber A Digital Equipment Corp. weber@star.enet.dec.com Mr. Edward A. Gardner A Digital Equipment Corp. Mr. David Wagner O Emulex Corp. d.wagner@emulex.com Mr. Wayne Rickard V Emulex Corp. Mr. I. Dal Allan P ENDL Mr. Howard Grill V Formation Mr. Mark Woithe O Future Domain Mr. Jeffrey L. P Hewlett Packard Co. Williams Mr. Kurt Chan A Hewlett Packard Co. Mr. Yoshihiko Yano O Hitachi Computer Products Mr. Gary R. Stephens A IBM Corp. gstephen@tucvm2.vnet.ibm. com Mr. John P. Scheible O IBM Corp. Scheible@sjevm5.vnet.ibm. com Mr. Larry Grasso S IBM Corp. Mr. Giles Frazier S IBM Corp. Mr. Wendell Lengefeld V IBM Corp. Mr. Kevin R. Pokorney O Intellistor, Inc. Mr. David Lawson P Interphase Corp. Mr. Lawrence J. Lamers P Maxtor Corp. 71540.2756@compuserve.com Mr. John Lohmeyer P NCR Corp. john.lohmeyer@ftcollinsco .ncr.com Mr. Bruce Anderson P NEC Technologies Inc. Mr. Stephen F. Heil P Panasonic Technologies, Inc Mr. Gene Milligan A Seagate Technology Mr. Paul W. Hanmann V Storage Concepts, Inc. Mr. Robert N. Snively P Sun Microsystems, Inc. Mr. John Moy O Tandem Computers moy_john@tandem.com Mr. Ricardo Dominguez P Texas Instruments ricardo.dominguez@hub.dsg .ti.com Mr. Alan Ebright P Toshiba America, Inc. Mr. Chuck Duquette P U.S. Design Mr. Doug Piper P Woven Electronics Mr. Greg Leonhardt P Zenith Data Systems 39 People Present Status (S) Key: P - Principal A - Alternate O - Observer L - Liaison V,S - Visitor 3. Approval of Agenda The agenda was developed at the meeting approved. 4. RAID Models In order to avoid preconceived notions coloring the discussions it was decided not to use the RAID level designations. Instead the following models of RAID behavior were used: Software RAID - The OS is in control of the array and is resposible for configuration an data reliability. Transparent RAID - The OS has no knowledge of the array during normal operations and is not responsible for data reliability. The array characteristics and statiscal information are available through maintenence software. Hybrid RAID - The OS is aware of the array and is responsible for the initial configuration but not the data reliability. The OS is notified of exception conditions so that a corrective action can be indicated. 5. Data Attributes Dal Allan reported that the RAID Advisory Board is trying to characterize RAIDS in terms of their data attributes to avoid the confusion being created by the various marketing folks claims of exotic RAID levels (e.g. 6, 7, 5-3, etc.). 6. Device Identification There is a need to get the essential information regarding the characteristics of each device in the array. This problem is compounded in multiple-level arrays and is associated with the physical addressability issue. One proposal is to require a that a logical unit zero exist on all devices at each level. The logical unit zero would provide the information on itself and the those devices directly attached to it. 7. Physical Addressibility The problems related to physical addressibility was a common thread throughout the disussions. Within the current SCSI there is not enough addressibilty to deal with a multiple-level array. There is also a notion that the logical unit number refers directly to an attached storage device. The SCSI-3 architecture model (SAM) does not impose this restriction per se. SAM has a 64-bit SCSI Device Identifier and a 64-bit Logical Unit Identifier. These identifier could be parsed to provide a hierarchecal addressing scheme. Bob Snively stated that FCP has both raid addressability and regular addressability as a feature so that the OS can perform normal operations and maintenance operations. 8. Reconstruction Operations What is needed for reconstruction operations are defined indicators, i.e., sense codes and log parameters. These issues will be dealt with under those items. One or more new commands may be needed for those environments wherein the array does not operate transparently. These commands could reconfigure an array (e.g., switch in a hot spare) or rebuild an array (e.g., replacement of a device in a stripped array). 9. Determining Devices State A mechanism is needed for an OS to poll the array and determine the current state and characteristics of the array. This could be similar to sense data except that an exception condition is not required for it to be generated. The information is not transient but is continuously updated. 10. INQUIRY data Extensions to INQUIRY data may be needed to identify RAID devices. 11. LOG SENSE data The descriptors in LOG SENSE need to be extended to report the statistics of the attached devices. A soft error reporting control is also needed. 12. REQUEST SENSE data New ASC and ASCQ codes are needed for RAIDs. Some of the new SCSI-3 codes may be appropriate. There is also a need for a code qualifier because hard errors should not occur. 13. Dynamic addition or deletion of capacity (lun) Some folks wanted to talk about the reconfiguring on the fly, adding or removing devices without a total reboot of the system. This is a thorny issue and is primarily a OS problem. 14. Replacement drives Must all drives be of the same type in an array?. If replaced device has equal or greater capacity the RAID should be able accept it. However this is an implementation decision by the RAID vendor and as such cannot be dictated. 15. Meeting Schedule The next meeting of the RAID Study Group is planned for January 11, 1993 at the Sheraton Hotel in Sunnyvale, CA. The meeting is expected to start at 3:00pm following the ATA working group meeting. Plan to stay till 7:00pm. 16. Adjournment The meeting was adjourned at 7:00 p.m. on Monday December 7, 1992.