bwf@ihuxz.ATT.COM (Fecht) (06/13/89)
Has the SCSI-II std group decided how many hosts and targets I can put on the bus? Are there any good trade articles on SCSI-II recently that I have missed? Thanks. Bill Fecht
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (06/13/89)
In article <4589@ihuxz.ATT.COM> bwf@ihuxz.ATT.COM (Fecht) writes: >Has the SCSI-II std group decided how many hosts and targets >I can put on the bus? Are there any good trade articles on >SCSI-II recently that I have missed? > >Thanks. >Bill Fecht The limit in SCSI-2 is eight. The committee has a proposal to extend the number of IDs to the bus width, which would mean that wide SCSI devices could have 16 or 32 addresses available. This has been designated a SCSI-3 proposal. This idea works by assigning lower priority to the additional IDs than for the first eight IDs. Thus new and old devices can be mixed on the same bus-- old devices obviously won't know how to talk to the new ones, however. There is a proposal which hasn't been formally submitted to the committee to define a true 16-bit SCSI. This would use one 68-conductor cable to support a 16-bit data path allowing 16 devices. The proposal is being developed by several committee members, including myself. I plan to write an article about it when it is a little further along. I've seen a few articles recently, but you would have to judge their "goodness" by your needs. I don't have a list of references, unfortunately. I would also appreciate a list of recent articles. John Lohmeyer, Chairman X3T9.2 (SCSI) J.Lohmeyer@Wichita.NCR.COM
kc@hprnd.HP.COM (Kurt Chan) (06/14/89)
> Has the SCSI-II std group decided how many hosts and targets > I can put on the bus? Sub-luns are out - leaving the same old 8 devices, 8 logical units per device. The maximum single-initiator system therefore consists of the initiator and up to 56 logical units spread across 7 targets (physical addresses). > Are there any good trade articles on SCSI-II recently that I have missed? The best stuff is coming from conferences: The annual SCSI Forum from a few months ago has a good set of technical and trade articles (4" binder) that you can have for a price: Joe Molina Technology Forums PO Box 1729 Minnetonka, MN 55345 612-934-1415 The IEEE Systems Design and Networks Conference (Mass Storage Session) held May 23-25, 1989 also has a nice set of technical papers (1" bound). About 80% of the papers deal with SCSI specifically. Call IEEE for information. Some (fairly) recent articles I've seen in trade rags: "The Quest for the Best", Andy Hospidor Mini-Micro Systems, 11/88 Testing considerations for SCSI disk drives and WORMs. "Disk Interfaces for the High End", Peter Aitken PC Tech Journal, 2/89 A comparison of SCSI, ESDI, and SCSI-2 written from a PC perspective. "Small Computer System Interface", Paul Q. Perlmutter Hewlett-Packard Journal, 10/88 A SCSI-1 tutorial with comparisons to HP-IB "SCSI Port on the VAXstation 2000" Victor Gould DEC Professional, March 1989 "Explore SCSI-2 Command Queues With Debugger" Ronald Berg and Mark Gordon Electronic Design, Feb 23, 1989 Mass Storage Supplement to PC Week April 10, 1989 Kurt Chan * Hewlett-Packard * Roseville Networks Division * 916-785-5621
terryk@xenna.Encore.COM (Terence Kelleher) (06/15/89)
In article <3170007@hprnd.HP.COM> kc@hprnd.HP.COM (Kurt Chan) writes: >> Has the SCSI-II std group decided how many hosts and targets >> I can put on the bus? > >Sub-luns are out - leaving the same old 8 devices, 8 logical units per device. >The maximum single-initiator system therefore consists of the initiator and >up to 56 logical units spread across 7 targets (physical addresses). > As of the August 1988 draft, revision 5, of SCSI-II, the EXTENDED IDENTIFY message was still supported. This allows up to 2048 (256 sub-logical units per LUN). Was this axed in a later draft? Not that I would miss it, but I am curious.
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (06/15/89)
In article <3170007@hprnd.HP.COM> kc@hprnd.HP.COM (Kurt Chan) writes: > ...stuff deleted... >> Are there any good trade articles on SCSI-II recently that I have missed? > > Kurt's list of recent articles deleted... I just received a copy of several related articles in the June 1989 issue of "Systems Integration" (formerly, "Mini-Micro Systems"). SCSI-2 is the cover story. Haven't read it yet, but it is likely to contain some good stuff. --- John Lohmeyer J.Lohmeyer@Wichita.NCR.COM NCR Corp. uunet!ncrlnk!ncrwic!entec!jlohmeye 3718 N. Rock Rd. Voice: 316-636-8703 Wichita, KS 67226 SCSI BBS 316-636-8703 300/1200/2400 24 hours
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (06/16/89)
In article <8399@xenna.Encore.COM> terryk@xenna.UUCP (Terence Kelleher) writes: >As of the August 1988 draft, revision 5, of SCSI-II, the EXTENDED >IDENTIFY message was still supported. This allows up to 2048 (256 >sub-logical units per LUN). Was this axed in a later draft? Not that >I would miss it, but I am curious. Yes, it was deleted in rev 7. The committee decided in December 1988 that this feature was poorly documented (when do you sent EXTENDED IDENTIFY in relation to other messages, etc.), few people (if any) actually used it, and it would be simpler to remove the message rather than document it correctly. We may have been wrong, but so far no one has proposed that it be reinstated... By the way, we could extend the number of logical units to 32 using reserved bits in the IDENTIFY message -- I like this way better.
bpenteco@hpdml93.HP.COM (Bob Pentecost) (06/16/89)
/ hpdml93:comp.periphs / terryk@xenna.Encore.COM (Terence Kelleher) / 2:42 pm Jun 14, 1989 / In article <3170007@hprnd.HP.COM> kc@hprnd.HP.COM (Kurt Chan) writes: >>> Has the SCSI-II std group decided how many hosts and targets >>> I can put on the bus? >> >>Sub-luns are out - leaving the same old 8 devices, 8 logical units per device. >>The maximum single-initiator system therefore consists of the initiator and >>up to 56 logical units spread across 7 targets (physical addresses). >> >As of the August 1988 draft, revision 5, of SCSI-II, the EXTENDED >IDENTIFY message was still supported. This allows up to 2048 (256 >sub-logical units per LUN). Was this axed in a later draft? Not that >I would miss it, but I am curious. >--------- Yes, the EXTENDED ID was deleted. I can't recall the exact reasons.
terryk@xenna.Encore.COM (Terence Kelleher) (06/16/89)
Seems like extending identify to allow 32 units is a good idea. Why are those bits reserved? Terry
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (07/22/89)
In article <8585@xenna.Encore.COM> terryk@xenna.UUCP (Terence Kelleher) writes: >Seems like extending identify to allow 32 units is a good idea. Why >are those bits reserved? > >Terry They were reserved in case we needed them! Like anything else that is reserved... 8-) Actually, we didn't need them at the time, so they were reserved for future standardization. The term "reserved" always means for future standardization (that is, until we (the committee) can think of a good purpose). It is not exactly just a matter of designating these bits for an extension to the logical unit field. We will also have to consider the impact of this change on such commands as COPY, which has some 3-bit logical unit fields in its segment descriptors. Nothing is easy. 8-( By the way, there used to be 3 reserved bits in the IDENTIFY message. One was used to specify a target routine instead of a logical unit (so a command can be directed to the target itself instead of a logical unit on the target). This is just to show that reserved bits do get used occasionally... --- -- John Lohmeyer J.Lohmeyer@Wichita.NCR.COM NCR Corp. uunet!ncrlnk!ncrwic!entec!jlohmeye 3718 N. Rock Rd. Voice: 316-636-8703 Wichita, KS 67226 SCSI BBS 316-636-8703 300/1200/2400 24 hours