Path: utzoo!attcan!uunet!tektronix!dadla!amadeus!jamesa From: jamesa@amadeus.LA.TEK.COM (James Akiyama) Newsgroups: comp.sys.ibm.pc Subject: Re: Leaving the hard disk on continuously Keywords: hard disk park Message-ID: <1373@amadeus.LA.TEK.COM> Date: 19 Jul 88 22:00:31 GMT References: <12184@agate.BERKELEY.EDU> Organization: Tektronix Inc., Beaverton, Or. Lines: 108 In article <12184@agate.BERKELEY.EDU>, cotner@jif.berkeley.edu (Carl Cotner) writes: > After some recent discussions and testimonials on the net, I've been > convinced that I should leave my PC and hard disk on 24 hours a day > to spare it the trauma of being powerred it on and off each time I > use the machine. > > Now I hear that I should park the heads of my disk whenever it is not in > use to prevent data corruption from a power surge. That sounds reasonable. > When I bought my Mountain HardDrive, the disk came with a head parking utility > for such a purpose. Now I'm wondering how do I UNPARK the hard disk to use it > again? > > As far as I know, I have to turn off the PC and turn it back on > in order to bring up the hard disk. However this this seems to defeat the > purpose of leaving the computer on continuously to save wear and tear on > the hard disk. Can anyone set me straight? Is there a utility to unpark a > parked disk? > > Thanks for any advice and input. > > cc > > PS: I'm also leaving my monitor on 24 hours a day, but with the brightness > switch dimmed to pitch black, operating under the same assumption that > transients during power on is not good for the monitor either. Is this > a reasonable assumption, or should I just switch the monitor off? First, a few notes about parking disk drive heads. The IBM PC/AT (not sure about PC's) contain the landing zone information as part of the information associated with the drive type table. The AT sends the fixed disk to the specified track when requested to park. Note that drives will vary in the method or need to park them. Most of the "voice coil" drives self-park upon power down and may not have a landing zone accessible during operation. Voice coil drives also tend to "lose" their track information when parked (otherwise it would be necessary to record track information in the parking zone--a no no). This requires them to perform a recalibrate and seek to zero upon "unparking"--both require a considerable amount of time. Note that some voice coil drives (e.g. Maxtor 1140) DO NOT SUPPORT BUFFERED SEEKS DURING THE INITIAL RECALIBRATE AND SEEK TO TRACK ZERO. In the case of the Maxtor drive, buffered seeks during the recalibrate and seek to zero may cause a high-speed crash at cylinder zero, damaging the head assembly. Note that Maxtor explicitly warns about this in their OEM manual. Because of this the initial "unparking" can be a relatively slow process. Other drives (the Miniscribe 8425 is an example) requires a specific cylinder to guarantee parking. In the case of the Miniscribe 8425 you must send the drive to cylinder 663 to park it--any other cylinder (higher or lower) will result in a seek error with the head generally going to a random location. Since many PC vendor's only worry about matching the drive type number to the disk's number of heads; number of cylinders and (hopefully) the write precompensation cylinder; and often times do not worry about the landing zone, a park utility may not actually park your drive. Some park utilities seek to the last cylinder location and consider the heads parked. This is not a valid park, in most cases. Although it is true that DOS does not use the last cylinder, other operating systems may (e.g. I believe Unix places the bad cylinder table there). Also the physical head is much larger than one track--a crash generally wipes out all neighboring cylinders as well (I've found tracks 50 cylinders away damaged). In fact, it is generally the cylinders under the edge (and not the center) of the head that are most likely to be damaged. Most drive manufacturers actually park the head many cylinders (30-50 cylinders) away from the data region. As far as leaving a PC on or turning it off, I believe that this depends a lot on the duration the equipment is going to be idle, the system configuration, and the design of the power supply. Power supplies which have soft start circuitry and a controlled output ramp during power-up, are much less stressful on the electronics (and themselves), then one's that do not. Also note that some PC's are not adequately cooled. These PC's may overheat if left on indefinitely, especially in a building where the air-conditioning is turned off during the weekends. Note that most PC's (at least IBM's) are designed to operate at temperatures where humans are comfortable. They are not designed to run in an unheated or excessively hot environment. In a PC, the fixed disk and monitor are the two most likely parts which would "wear out" if left on. In the case of the monitor, this can be helped by either turning down the brightness or using a blanking utility. Note that this prevents the phospher from becoming "burned" but does not eliminate wear on the cathode gun. Over a long term, this may result in the CRT becoming dim as the tube begins to "wear out". In the fixed disk the normal idle wear process is generally associated with the spindle motor and spindle bearing. I have found at work, that drives will tend to wear out in about 1 to 2 years if left on continuously (although these figures are improving with the newer drives). As far as my personal PC's, I generally power them up when I need them, then power them back down before going to bed. Since I am now looking into being able to do remote logins I may have to change this a bit. Right now I'm looking at building some circuitry which allows remotely powering down/up the fixed disk under software control. As far as the monitor, I will probably power it down when I'm not present (I never did trust any unattended CRT). The rest of the PC will remain powered up constantly. I'm not entirely sure this is the best approach, but, in my opinion, it is at least reasonable. Oh, I should mention that these opinions are my own and not part of Tektronix. As far as the fixed disk stuff, I'm fairly sure most of it is accurate since part of my job description at Tek is to evaluate fixed disk for our product lines. We generally perform extensive burn-in and environmental testing before qualifying anything; unfortunately, most of the resulting data is considered company confidential. Hope this helps. James E. Akiyama jamesa@amadeus.LA.TEK.COM UUCP: ....!tektronix!amadeus!jamesa ARPA: jamesa%amadeus.LA.TEK.COM@RELAY.CS.NET