From mboxrd@z Thu Jan 1 00:00:00 1970 From: clemc at ccc.com (Clem Cole) Date: Wed, 3 Jul 2019 14:38:44 -0400 Subject: [COFF] Disk Technology was [Simh] Which PDP-11 to choose In-Reply-To: References: <6e8cdcbf-7183-1082-0437-403a6f3b2994@gmail.com> <25855953-9505-481C-A0E2-1AAD53B7BEC5@ccc.com> <28FA3347-B880-400A-B606-1240C83FA867@ccc.com> <005401d5300f$74588a60$5d099f20$@twsoft.co.uk> <20190701141124.GP1912@mcvoy.com> <00c901d530db$cf7b40f0$6e71c2d0$@twsoft.co.uk> Message-ID: John - not much there: Introduction to redundant arrays of inexpensive disks (RAID) >From Page 112, Bottom of 1st column continued to the top of second the text is: While the magnetic disk industry has made little progress in improving speed of disks, it has significantly reduced the size of disks. The personal computer industry has created a market for 5.25 and 3.5 inch drives, reducing the cost per disk system as well as the traditional lowering of cost per megabyte. Table I below compares the top-of-the-line IBM 3380 model AK4 mainframe disk, Fujitsu M2361A "Super Eagle" minicomputer disk, Impress/CDC Wren-IV workstation disk, and the Conner Peripherals CP 3100 personal computer disk. >From Page 116, Second column the text is; One problem that several magnetic disk manufacturers have mentioned is what we would call the "Pinto Effect;" a mistake is made in manufacturing process that is so disastrous that the disk manufacturer will recall all affected disks and replace them. The common theme is that the mistake is uncovered after the disks have been in the field for several months and the disks all fail within a short time of one another. One example was a manufacturer who glued together the two halves of an head-disk assembly, with this glue dissolving after the disks had been in the field for 18 months. Another example was that a new bateriacide used in an air filter interacted with the disk surface so that many failures occurred six months later. A common cause of the Pinto Effect is that a supplier will change some component as a cost cutting measure without notifying the disk manufacturer, and disastrous consequences occur due to unforeseen interactions. Although we desperately need real data on disk failures, we are performing studies of using models to estimate the impact of the Pinto Effect on RAID reliability. On Wed, Jul 3, 2019 at 2:02 PM John P. Linderman wrote: > The best I could find googling *fujitsu super eagle "glue"* was > > Introduction to redundant arrays of inexpensive disks ... - IEEE Xplore > https://ieeexplore.ieee.org/iel2/231/7454/00301912.pdf > > by DA Patterson - ‎1989 - ‎Cited by 324 > - > ‎Related articles > > computers, the Fujitsu M2361A "Super Eagle" disk for minicomputers, ..... > assembly, with this gluedissolving after the disks had been in the field > for 18 months. > > Tantalizing, but I couldn't dig further, perhaps because I'm not a member > of IEEE. > > On Tue, Jul 2, 2019 at 5:34 PM Rico Pajarola wrote: > >> if you still have that story, I'd love to hear it. A quick search didn't >> turn up anything except that Super Eagles is a Nigerian football team. >> >> >> On Tue, Jul 2, 2019 at 12:59 PM John P. Linderman >> wrote: >> >>> I don't have authoritative info on the cause, I'm just repeating what I >>> heard. A highly viscous substance like glue would explain why it took so >>> long to fail. In any event, it was nasty. Worked perfectly long enough to >>> build confidence, then failed spectacularly. It was widespread. I entered a >>> "Sysadmin Horror Story" contest at a USENIX (San Diego?), and won with a >>> "short story" entry: *Supereagles*. I still have the shark's tooth >>> trophy. >>> >>> On Tue, Jul 2, 2019 at 9:41 AM Tim Wilkinson wrote: >>> >>>> >>>> >>>> Interesting that it was the platter bonding. The explanation SI gave us >>>> (They sold us the super Eagles along with their controller) was that it was >>>> a lubricant. So I had assumed a bearing seal fail. >>>> >>>> >>>> >>>> Anyway after about 4 swap outs and a lot of lifting they lasted a >>>> further 15 years until we knocked down the office with the original 750 and >>>> its big brother an 8810 still in the computer room as all the resellers >>>> wanted certificates of continuing maintenance that would have cost more >>>> than they were willing to pay for those vaxs. >>>> >>>> >>>> >>>> *From:* John P. Linderman [mailto:jpl.jpl at gmail.com] >>>> *Sent:* 02 July 2019 12:47 >>>> *To:* Larry McVoy >>>> *Cc:* Clem Cole ; Patrick Finnegan < >>>> pat at computer-refuge.org>; COFF ; Tim Wilkinson < >>>> tjw at twsoft.co.uk> >>>> *Subject:* Re: [COFF] Disk Technology was [Simh] Which PDP-11 to choose >>>> >>>> >>>> >>>> There were eagles, and then there were super-eagles. Our experience >>>> with eagles was great, and we were eager to try the (larger) super-eagles. >>>> We soaked them for a month or so, then put them into production use. >>>> Whereupon, they started dropping like flies. It turns out the glue they >>>> used to attach the platters to the spindle slowly crept out over time, >>>> eventually coming to grief with a read/write head. This experience was >>>> wide-spread, and seriously damaged Fujitsu's reputation. >>>> >>>> >>>> >>>> On Mon, Jul 1, 2019 at 10:11 AM Larry McVoy wrote: >>>> >>>> On Mon, Jul 01, 2019 at 09:49:42AM -0400, Clem Cole wrote: >>>> > An Eagle or Eagle-II was a whole lot lighter (and physically >>>> smaller) than >>>> > an RP06 or RP07 (or an RM series drive for that matter). It is >>>> interesting >>>> > to hear you had problems with the Eagles. They were generally >>>> considered >>>> > the best/most reliable of the day. The SI controller on the Vax was >>>> less >>>> > so, although many of us in the UNIX community used them. >>>> >>>> We ran Eagles on the Masscomps we had at Geophysics. Nothing but good >>>> things to say about those drives. >>>> _______________________________________________ >>>> COFF mailing list >>>> COFF at minnie.tuhs.org >>>> https://minnie.tuhs.org/cgi-bin/mailman/listinfo/coff >>>> >>>> >>>> >>>> Virus-free. >>>> www.avast.com >>>> >>>> <#m_2142356106908993712_m_1751084755802734138_m_-5102461650516400670_m_-7814325232288781299_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2> >>>> >>> _______________________________________________ >>> COFF mailing list >>> COFF at minnie.tuhs.org >>> https://minnie.tuhs.org/cgi-bin/mailman/listinfo/coff >>> >> _______________________________________________ > COFF mailing list > COFF at minnie.tuhs.org > https://minnie.tuhs.org/cgi-bin/mailman/listinfo/coff > -------------- next part -------------- An HTML attachment was scrubbed... URL: