Re: Hardware DASD design question.
In a message dated 6/22/2005 7:23:43 A.M. Central Daylight Time, [EMAIL PROTECTED] writes: Only certain drives. The S/360 DASD and the original S/370 DASD did not have skip displacement, although they did have alternate tracks. IIRC, the 3330 was the first DASD that had skip displacements. Before leaving the factory, a new disk was subjected to rigorous testing in a very sensitive environment. If a bad spot was found on a track that was small enough to be covered by skip defects, one or more defective spots were indicated on the track by recording their offset from the track's index point in special 2-byte fields called skip displacements. The controller had logic to look for this info on the track and, if found, it would automatically skip over the defective spots thus marked. If a defective spot was too large to be covered by all possible skip defects on one track, then an alternate track might be assigned or, if the platter had too many such tracks, the entire platter would be rejected. All skip defects found and assigned in the factory were recorded in the diagnostic tracks area, accessible only to authorized programs. The max number of defects per track grew larger over the years as recording densities got higher. I think the 3330 had one, 3340s and 3350s had three, and 3375s, 3380s, and 3390s allowed up to seven skip defects per track. But all that is arcane virtual history now, as no new SLED 3390s have been manufactured by any vendor for many years. Only RAIDed FBAs are produced now. Bill Fairchild -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Hardware DASD design question.
In [EMAIL PROTECTED], on 06/22/2005 at 08:59 AM, Bill Fairchild [EMAIL PROTECTED] said: IIRC, the 3330 was the first DASD that had skip displacements. My recollection is that the 3340, 3344 and 3350 were the first. I'll have to see whether I still have my 3330 manuals. -- Shmuel (Seymour J.) Metz, SysProg and JOAT ISO position; see http://patriot.net/~shmuel/resume/brief.html We don't care. We don't have to care, we're Congress. (S877: The Shut up and Eat Your spam act of 2003) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Hardware DASD design question.
In a message dated 6/22/2005 1:37:35 P.M. Central Daylight Time, [EMAIL PROTECTED] writes: My recollection is that the 3340, 3344 and 3350 were the first. I'll have to see whether I still have my 3330 manuals. Could well be. I hesitated on saying 3330. I think the 3340 and 3344 had 3 SDs and the 3350 had 5. Too many decades ago to remember clearly. They always seemed to go up by 2 whenever they increased. 3375/3380/3390s had the most - 7 SDs. The way these worked was pretty clever. The first thing ever permanently recorded on a track is the Home Address. Once skip displacements were invented, the SDs were saved in a normally unreadable part of the Home Address. To read them in or write them back out, you had to use the Read/Write Special Home Address CCW. This read in 27 bytes on the 3375 and 28 on 3380/3390, 14 bytes of which were 7 different 2-byte SDs. Then when you wrote R0 after the HA, the controller copied the 7 SDs that it found in the HA into an unreadable part of R0 (a glorified count field, IIRC). Then whenever you wrote the first record after R0, the controller copied the 7 SDs into an unreadable part of this new record. Thus the same 7 SDs were propagated into each new record written on the track, and no matter where you were on the track when you first established orientation, the controller would have all 7 SDs available the next time it sensed any count field. Bill Fairchild -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Hardware DASD design question.
In a message dated 6/22/2005 2:46:21 P.M. Central Standard Time, [EMAIL PROTECTED] writes: couple of devices/controller types, and is no longer documented. It would let you read ALL the bytes on the track, including the inter-record gaps and all these SDs, inter alia. But what you read in was not much fun to decipher by hand and eyeball. Under MFT there was bug in OPEN that positioned you at CC HH 00 00 if you did a rewind before open. The IBM PSR poo-pooed the observation with something like-NFWIMLT! Next day came to work to find his test GIS pack relabeld with his INITIALs iiiYTI(yes there is!) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html