mc68020@gilsys.UUCP (Thomas J Keller) (08/30/87)
In general, it is my considered opinion that leaving a hard-disk based
system powered up is a good thing, as it seems to reduce hardware failures.
An interesting situation arises, however, when one is using consumer grade
winchester drives with low MTBF (mean time between failure) ratings.
Take the Seagate ST-225 piece of trash, for example. A very popular
drive, largely due to its low price, this drive has a 4000 hour MTBF. Now,
a little simple arithmetic will reveal that 4000 hours is sightly less than
six months. Of course, there is no guarantee that the drive *WILL* fail in
4000 hours, only that the statistical mean over the sample population used
in this test was 4000 hours.
Nevertheless, the question arises: which is the better approach with such
a low-quality drive? To leave it powered up, thereby frittering away a LARGE
percentage of its expected lifespan, or to power-cycle the system, thereby
stressing the drive (and the rest of the system) with THOSE problems?
I honestly cannot decide which makes more sense. What makes *REAL* sense
is to avoid consumer-grade disk drives like the plague, and stick to the
commercial-grade, higher quality devices. Sometimes immediate economics
make such a move impossible, however. What to do, what to do?
Comments or suggestions via email, please. Flames and smart-a** remarks
> /dev/null.
--
Tom Keller
VOICE : + 1 707 575 9493
UUCP : {ihnp4,ames,sun,amdahl,lll-crg,pyramid}!ptsfa!gilsys!mc68020
root@hnsurg3.UUCP (09/01/87)
+---- mc68020@gilsys.UUCP (Thomas J Keller) writes in <1118@gilsys.UUCP> ---- | | Take the Seagate ST-225 piece of trash, for example. A very popular | | (text deleted.....) | | I honestly cannot decide which makes more sense. What makes *REAL* sense | is to avoid consumer-grade disk drives like the plague, and stick to the | commercial-grade, higher quality devices. Sometimes immediate economics | +-------------- The net can really USE an on-going REFERENCE section, perhaps in the Archives, for such information as which HD's to avoid, etc.... especially as a source to which NEW net-users can be referred, AND for subjects prone to CYCLIC REPETITION of questions-answers as happens in a number of areas (PC *AND* big-box hard disks; Endian wars ; (m)alloc; What is a Byte; - save the comments by the experienced folks - and a number of other REPEAT subjects)... Could save a lot of traffic when the same questions appear every 3-4 months... Comments - Harris, Gwyn, Adams, Horton, Spencer, Spafford, et allei ? -- Don Lawrence {decvax,cbatt,cbosgd}!hal!ncoast!hnsurg3!don (Generic disclaimer... ) seismo!uunet!hnsurg3!don -- Don Lawrence {decvax,cbatt,cbosgd}!hal!ncoast!hnsurg3!don (Generic disclaimer... ) seismo!uunet!hnsurg3!don
gk@kksys.UUCP (Greg Kemnitz) (09/05/87)
In article <1118@gilsys.UUCP> mc68020@gilsys.UUCP (Thomas J Keller) writes: <....Take the Seagate ST-225 piece of trash, for example. A very popular <drive, largely due to its low price, this drive has a 4000 hour MTBF.... < <Tom Keller <VOICE : + 1 707 575 9493 <UUCP : {ihnp4,ames,sun,amdahl,lll-crg,pyramid}!ptsfa!gilsys!mc68020 Actually, from the info I have it appears the ST-225 has a very respectable 20,000 hour mtbf.... I am including a copy of part of an earlier posting which may help to clarify this: <Newsgroups: comp.sys.ibm.pc,comp.misc <Subject: Re: Leaving AT on overnight (good/bad) <References: <1246@bloom-beacon.MIT.EDU> <768@custom.UUCP> <2348@trwrb.UUCP> <410@polyslo.UUCP> <Reply-To: gk@kksys.UUCP (Greg Kemnitz) <Distribution: usa <Organization: K and K Systems, Minneapolis < <In article <410@polyslo.UUCP> npollack@polyslo.UUCP (Neal Pollack) writes: <>> Unfortunately, many of the winchesters used in the PC and <>> small-UNIX machines only have a MTBF rate of ~3000 hours. A <>> -- <>> Marc A. Ries <>NOT TRUE; <> <>While I too have seen a few small winchesters die at around 3000 hours, <>most are rated for 5 years life span. If you look at the OEM hardware <>manual for seagate drives, of which I own one, the entire series is <>rated at a MTBF (failure time) of approx 25,000 hours. That is more than <>3 or 4 years at 24 hours a day. Turning it on or off each day during <>this same time would provide more than 1000 takeoffs and landing for <>potential crashes, near misses, presidential fly bys, etc. :-) < <I recently received a brochure from Seagate which gives specs on their <current drive lines... It shows the ST125, 138, 213, 225 and 251 as <having a MTBF of 20,000 hours, and the ST4026, 4038, 4051 and 4096 as <having a MTBF of 15,000 hours. At 20,000 hours, this would be about 2 <years and four months... at 15,000 hours it works out to a bit under <one year and nine months, assuming 24 hour a day operation. < <They -do- list the drives as having a service life of five years, and <a mean time to repair (MTTR) of thirty minutes. < <>| Neal S. Pollack | ...!ihnp4!csun!polyslo!npollack | <>| Computer Systems Lab | ...!{csustan,csun,sdsu}!polyslo!npollack | <>| Cal Poly State Univ. | ...!ucbvax!voder!polyslo!npollack | <>| San Luis Obispo, CA 93407 | Voice: (805)-546-2147 | < < <Greg Kemnitz | ...!amdahl \ <K and K Systems | ...!ihnp4 !meccts!kksys!gk <P.O. Box 41804 | ...!rutgers/ <Plymouth, MN 55441 | AT&T and clones: (612)475-1527 If anyone would like a copy of the Seagate drive brochure, I have several extras and will gladly forward one to you via USsnail. As my .signature file is included above, I will not repeat it...
mlinar@poisson.usc.edu (Mitch Mlinar) (09/19/87)
>In article <1118@gilsys.UUCP> mc68020@gilsys.UUCP (Thomas J Keller) writes: ><....Take the Seagate ST-225 piece of trash, for example. A very popular ><drive, largely due to its low price, this drive has a 4000 hour MTBF.... > >Actually, from the info I have it appears the ST-225 has a very >respectable 20,000 hour mtbf.... I am including a copy of part of an earlier >posting which may help to clarify this: > ><>> Unfortunately, many of the winchesters used in the PC and ><>> small-UNIX machines only have a MTBF rate of ~3000 hours. A > ><>While I too have seen a few small winchesters die at around 3000 hours, ><>most are rated for 5 years life span. If you look at the OEM hardware ><>manual for seagate drives, of which I own one, the entire series is ><>rated at a MTBF (failure time) of approx 25,000 hours. That is more than >< ><I recently received a brochure from Seagate which gives specs on their ><current drive lines... It shows the ST125, 138, 213, 225 and 251 as ><having a MTBF of 20,000 hours, and the ST4026, 4038, 4051 and 4096 as ><having a MTBF of 15,000 hours. At 20,000 hours, this would be about 2 ><years and four months... at 15,000 hours it works out to a bit under ><one year and nine months, assuming 24 hour a day operation. >< ><They -do- list the drives as having a service life of five years, and ><a mean time to repair (MTTR) of thirty minutes. Well, to add to all this, you must understand HOW these numbers are determined, which is usually by test of a small group under CONTROLLED conditions. As you have heard me write before, I have tested 100's of drives for both lifetime and controller card driver debugging. This put me in close contact (still am) with several dealers who collectively push out 100+ hard drives a month for micros. The following position is based on this experience. I have YET to see a hard drive last 3 years in an IBM PC or any other micro, limited cooling and all. (I am talking about 3 years of USE, not "I use my machine once or twice a week/month and it has lasted 25 years....") This can range from as low as one year to as high as 2.5 years (excluding infant mortality). Drives include CDC, Tandon, CMI, Shugart, Maxtor, Seagate, Rodime, MicroSci, and a bunch that you may not have heard of. (If someone wants the FULL boring list of manufacturers, let me know.) Yes, SOME are better than others, in the MICROCOMPUTER environment. Dealers reported Tandons and Seagates being returned as fast as they could order replacements whereas Rodime and Maxtor were rarely returned. (Not to slight the others, but these are the ones I remember.) Bearing seizures were the most common followed by media failures. Plated media virtually eliminates media failures, but at a premium in price. In short, for the current consumer market drives, assume about 2 years of moderate use before failure. If you use your drive once or twice a week, you are better off to shut it down rather than spin it. My rule of thumb is that use within 24 hours determines whether to shutdown or leave spinning. It is interesting to note that nearly all failures for home computers are bearing seizures; these drives spend only a fraction of their life accessing media and the rest just merrily spinning away. Finally, MTTR, by many manufacturers, has come to mean Mean Time to Replace. Yes, it only takes 30 minutes or less to replace the hard drive. However, if you don't have current backups or $$$ for a new one, this is meaningless. -Mitch
corwin@cdp.UUCP (09/23/87)
I can highly recommend Micropolis drives as well. I have used quite a number and have never had one failure. They publish an MTBF of over 55,000 hours on their 1320 series (85 MB), and MTBF of over 32,000 hours on the 1350 and 1370 series (135 MB and 170 MB). I've only used the 1325 models. Your mileage may vary... Corwin Nichols ...hplabs!cdp!corwin
mc68020@gilsys.UUCP (Thomas J Keller) (10/02/87)
In article <4464@oberon.USC.EDU>, mlinar@poisson.usc.edu (Mitch Mlinar) writes: > > I have YET to see a hard drive last 3 years in an IBM PC or any other micro, > limited cooling and all. (I am talking about 3 years of USE, not "I use my > machine once or twice a week/month and it has lasted 25 years....") > This can range from as low as one year to as high as 2.5 years (excluding > infant mortality). Drives include CDC, Tandon, CMI, Shugart, Maxtor, Seagate, > Rodime, MicroSci, and a bunch that you may not have heard of. Well, since I started this brouhaha, I would like to comment on this last item. I am runing a Tandy MOdel 16A (upgraded to a MOdel 6000) with a 12 Mbyte (formatted) Tandon 603S 5.25" winnie that has been in continuous service for over 5 years now. BY continuous, I mean that in the past 62 months, this drive has spent a sum total of perhaps 48 hours *NOT* spinning. As regards Seagate ST2[23][58] drives, I don't give a hangdogged DAMN what Seagate CLAIMS for them, these drives are low-cost consumer trash that will not likely provide more than 2 years of service at best. In one situation, a comany I was working with had over 60% of ST225 drives fail within the first 6 weeks. This was over a sample of approximately 175 drives, and a 6 month time frame. Seagate INSISTED there was nothing wrong with their drives. For my money, I wouldn't recommend ANY Seagate product to ANYONE except as a boat anchor or a door stop. -- Tom Keller VOICE : + 1 707 575 9493 UUCP : {ihnp4,ames,sun,amdahl,lll-crg,pyramid}!ptsfa!gilsys!mc68020