leemc@csri.toronto.edu (Matthew Lee) (01/04/90)
I have been saddled with the privilege of investigating the following
proposal:
We are currently running a Novell network with two servers and 40 NEC 386SX
*workstations* without hard disks. It has been suggested that since this
system is only in use from 8:00 a.m. to midnight that it be powered down for
the 8 hours at night that it is not in use, the reasoning being that by reducing
the period that the system is powered on by 1/3, the life of the machines will
be prolonged by a significant amount. The system would probably be powered on
and off by some sort of timer installed in the breaker panel since there would
not be anybody around at the required times.
My question is: Is this really a worthwhile endeavour? I know this subject has
been discussed here before particularily with reference to the effects of
powering on/off on hard disks. Hence I suspect that the servers would be best
left on 24 hrs/day. However, since the 386's are hard diskLESS I have no idea
what the relative merits of this proposal are.
Anybody out there who has done something similar or is in a position to comment
knowledgably on the subject? Responses by e-mail would be welcomed and
appreciated. I will summarize to the net if I get something worthwhile.
Matthew Lee
leemc@csri.toronto.edumrichey@orion.oac.uci.edu (Mike Richey) (01/04/90)
In article <1990Jan3.122707.1625@jarvis.csri.toronto.edu> leemc@csri.toronto.edu (Matthew Lee) writes: >I have been saddled with the privilege of investigating the following >proposal: > >We are currently running a Novell network with two servers and 40 NEC 386SX >*workstations* without hard disks. It has been suggested that since this >system is only in use from 8:00 a.m. to midnight that it be powered down for >the 8 hours at night that it is not in use, the reasoning being that by reducing >the period that the system is powered on by 1/3, the life of the machines will >be prolonged by a significant amount. The system would probably be powered on >and off by some sort of timer installed in the breaker panel since there would >not be anybody around at the required times. Powering up equipment is when most electrical damage occurs. When the power is turned on, all of the electronic circuits behave like a short. This only happens for a brief period of time, but the current surge is large. Most file servers are left powered on 24 hours a day in order to avoid the damage that may occur during power up sequences. Take a look around at most multi user computer system shops. The systems are never powered down unless there is a problem that requires it. The cost of electricity may be small compared to problems that arise from power down/up sequences. I know of servers (Novell NW) that have been on for months at a time, and are years old that rarely are powered off. I wouldn't even down the server during those eight hours, (MHO). > >My question is: Is this really a worthwhile endeavour? I know this subject has >been discussed here before particularily with reference to the effects of >powering on/off on hard disks. Hence I suspect that the servers would be best >left on 24 hrs/day. However, since the 386's are hard diskLESS I have no idea >what the relative merits of this proposal are. Well, I leave mine on 24 hours a day, except for long weekends, holidays. Again, my opinion. Michael S. Richey Internet: mrichey@orion.oac.uci.edu Bitnet: MRichey@UCI CompuServe: 71650,3132 Voice: (714) 856-8374 University of California, Irvine Network and Telecommunications Services 342 Computer Science Irvine, CA 92717
poffen@molehill (Russ Poffenberger) (01/05/90)
In article <1990Jan3.122707.1625@jarvis.csri.toronto.edu> leemc@csri.toronto.edu (Matthew Lee) writes: >I have been saddled with the privilege of investigating the following >proposal: > >We are currently running a Novell network with two servers and 40 NEC 386SX >*workstations* without hard disks. It has been suggested that since this >system is only in use from 8:00 a.m. to midnight that it be powered down for >the 8 hours at night that it is not in use, the reasoning being that by reducing >the period that the system is powered on by 1/3, the life of the machines will >be prolonged by a significant amount. The system would probably be powered on >and off by some sort of timer installed in the breaker panel since there would >not be anybody around at the required times. > >My question is: Is this really a worthwhile endeavour? I know this subject has >been discussed here before particularily with reference to the effects of >powering on/off on hard disks. Hence I suspect that the servers would be best >left on 24 hrs/day. However, since the 386's are hard diskLESS I have no idea >what the relative merits of this proposal are. > >Anybody out there who has done something similar or is in a position to comment >knowledgably on the subject? Responses by e-mail would be welcomed and >appreciated. I will summarize to the net if I get something worthwhile. I know other people have responded, but let me re-iterate. There is MUCH more harm than good by powering off at night and back on in the morning. Although the "power surge" mentioned is not necessarily the biggest culprit, but the thermal cycling that causes mechanical stress on the components. For example, it is a well known fact that a light bulb will last longer (on time) if it is left on, rather than turning it off and on. Russ Poffenberger DOMAIN: poffen@sj.ate.slb.com Schlumberger Technologies UUCP: {uunet,decwrl,amdahl}!sjsca4!poffen 1601 Technology Drive CIS: 72401,276 San Jose, Ca. 95110 (408)437-5254
akcs.amparsonjr@vpnet.UUCP (Anthony M. Parson, Jr.) (01/19/90)
Found it interesting to read the question again. I turn my home computer on and off as needed, off every night. However, tho all indications are that this is bad, I haven't had problems. At work, we just received an "official" policy change, that ALL personal computers, printers, etc are to be turned off at night, to save electrical energy. I guess that will test a number of our PC's! (the terminals connected to a network will remain on all the time, but not printers connected to those terminals). ---------------------------------------- | STANDARD DISCLAIMER: | | "... on the other hand, some very | UUCP: akcs.amparsonjr@vpnet.uucp | knowledgable people may justifiably | | disagree with my comments, in whole | C-serve: 73336,1667 | or in part ... " | ----------------------------------------