Leaving Dell Dimension 8300 running 24/7 ...?

[Thomas G. Marshall]

Ok, but Steve was pointing out what he saw as a bottom line, which was that
for whatever reason the systems left on seemed to have far fewer crashes.
/Regardless/ of the underlying reasons, Steve makes a cogent argument when
armed with a sampling of several computers, no?

IOW, are you saying that there /must/ be some other explanation for Steve's
observations? If so, what might it be?


w_tom coughed up:

 

[Paul Knudsen]

Questions:

1. Is the 8300 cooled enough or otherwise built for staying on? I'll of
course use power options to shutdown unnecessary things and brown down
perhaps the motherboard or something. I'll have to learn more about
this---I'm half ignorant on all things power control except for hibernation.

My wife leaves her 8300 on all the time. No problems so far. I can
see the back of it from here and the network activity light flashes
occasionally!

Seriously, they go into standby mode and use very little power so
cooling should not be a problem.

 

[Julian]

I am considering leaving the system on 24/7 and establishing a daily viral
sweep.

Why leave a computer on 24/7 if it isn't a server? We all have to start
to look for ways to curb excessive energy use and prevent global
warming, and one of the most painless ways to do this, it seems to me,
is to turn equipment off when you aren't using it.

 

[Julian]

6. The debate about leaving a computer powered up 24/7 or powered down when not
in use centers around wear-and-tear. Those who prefer to leave a computer up
24/7 point to the wear-and-tear on system electronics due to the zero-to-60
effect of a sudden surge of current after a total absence of power. Those who
prefer to power down a computer point to the wear-and-tear of the bearings on
rotating motors, notably fans and the hard drives. For me, the hard drive AND
its contents are the most important part of my system, even with regular
backups. I can always replace a blown power supply, motherboard, CD-ROM drive,
memory, or ANY other part of a computer. But I cannot replace the data. So I
am in the power-it-down camp... Ben Myers

Same here, also for environmental reasons. I have a 7 year old Dell
Dimension XP 400 which was in daily use by me until a few months ago,
when it became my wife's computer. It has been used on average 10 hours
a day 5 days a week for all that time, always switched off at night, and
has never suffered a single hardware failure.

 

[Julian]

Seriously, they go into standby mode and use very little power so
cooling should not be a problem.

They still use more power than when actually shut down. With power
saving enabled the HD and monitor power off when the system isn't being
used, so the wear and tear aspect of starting and stopping is still present.

 

[Sparky]

1. The system is perfectly suited for staying on 24/7. Whether it's better
to keep it on vs. powering it off is an ongoing (decades, now) debate with
no clear winner... Do what is best for your situation.

This "leave the computer ON" discussion goes back much farther than Bill
Gates dreaming of being a billionaire. I worked for IBM 1965-70 and it
was common wisdom then that the most likely time for a [mainframe]
system failure was being when being powered up. Many shops at this time
who didn't run 24/7 left their S/360's on all weekend to avoid a power
on situation Monday morning.

 

[Ben Myers]

And we powered down our GE-225 computer every night back in the '60s, unless I
was working around the clock doing some programming. Of course, it DID consume
a bit of electricity. Less powerful than an original IBM XT, but with
processing unit the size of 4 large refirgerators, hard disk pizza oven, and a
bank of tape drives, all on maybe 1000 square feet of raised flooring. Of
course, there were no power saving options back then... Ben Myers

1. The system is perfectly suited for staying on 24/7. Whether it's better
to keep it on vs. powering it off is an ongoing (decades, now) debate with
no clear winner... Do what is best for your situation.

This "leave the computer ON" discussion goes back much farther than Bill
Gates dreaming of being a billionaire. I worked for IBM 1965-70 and it
was common wisdom then that the most likely time for a [mainframe]
system failure was being when being powered up. Many shops at this time
who didn't run 24/7 left their S/360's on all weekend to avoid a power
on situation Monday morning.

 

[Thomas G. Marshall]

Julian coughed up:

Why leave a computer on 24/7 if it isn't a server? We all have to
start to look for ways to curb excessive energy use and prevent global
warming, and one of the most painless ways to do this, it seems to me,
is to turn equipment off when you aren't using it.

Because otherwise it's too freaking cold in my computer room.

No, seriously, the reason for leaving it on 24/7 is that I *don't* want the
AV scan to run while I'm actually using it. Thus it seems that the best and
easiest way to handle that is to run the AV scan late or at least when I'm
done with it.

Unfortunately, it then becomes out of sight and mind, which means that I'll
never get around to checking for its completion and turning off the system.
And checking for that is a pain in the ass anyway.

Which brings me back to this equation:

Scanning every day == System always on

I don't like the notion very much, but it is reality. At least in my puny
universe.

--
"This creature is called a vampire. To kill it requires a stake
through its heart." "I shall drive my staff deep into its rump."
"No no, this creature is from a dimension where the heart is in the
chest." "....Disgusting."

Demons discussing "Angel", a good vampire from our dimension visiting
theirs.

 

[Julian]

No, seriously, the reason for leaving it on 24/7 is that I *don't* want the
AV scan to run while I'm actually using it. Thus it seems that the best and
easiest way to handle that is to run the AV scan late or at least when I'm
done with it.

Unfortunately, it then becomes out of sight and mind, which means that I'll
never get around to checking for its completion and turning off the system.
And checking for that is a pain in the ass anyway.

Which brings me back to this equation:

Scanning every day == System always on

I don't like the notion very much, but it is reality. At least in my puny
universe.

You could write a bit of VB script to do the scan and then shut down the
computer when it has finished.

 

[w_tom]

The computer left on 24/7 was finally shut off. Then it
would not power up. This computer was now considered failed
due to being 'power cycled'. But it should have remained a
prime example of why 24/7 is so destructive. In the meantime,
once we first made the distinction, then suddenly computers
left on 24/7 were failing more often.

We see this with computers, TVs, radios, and so many other
electronics (not to be confuses with high tech, custom
designed circuits designed to maximize other parameters - ie
Deep Space Network). If leaving computers on 24/7 was so much
better, then we must also leave on TVs and radios 24/7. One
cannot have it both ways. Either Steve must also recommend
leaving TV and radios always on, or he has misrepresented the
data by doing subjective observations.

Bottom line: once we confronted how these people were
'collecting' the data, then suddenly numerous computers listed
as 'failed due to power cycling' were transferred to the
column that said 'failed due to 24/7 wear'.

Provided is how one must first learn what caused the damage
- forensic analysis often at the electronic component level.
Provided are numbers taken from manufacturer data sheets that
demonstrate why power cycling is not so destructive. Provided
is how those who recommend 24/7 fail the properly collect
their data - and then make subjective conclusions.

And then the contrarian example. If leaving computer on
24/7 is better, then leave your TVs and radios on also.

In parallel example, the many will also declare that power
cycling is destructive to incandescent lamps. They will use
the same subjective analysis. For example, light bulb burns
out most often when powered on. Again, they failed to first
learn why light bulbs burn out. Hours of operation and some
other parameters. Power cycling does not appear on a list of
reasons for light bulb failure. And yet so many using the
same subjective data will say otherwise. Those same reasons
used to promote 24/7 computer operation will also declare
power cycling destructive to light bulbs.

I one example, the fan failed prematurely. Then when
powered up, the fan inside that power supply did not start
spinning. Therefore a PSU component failed. The repairman
then blamed power supply failure on power up surge. In
reality, the failure was due to a fan with too many hours of
operation and a mislocated hall effect sensor. Where do you
put this failure? Due to power up? Or due to too many hours
of operation? Or due to manufacturing defect? Without the
details, one might immediately cite this as an example of why
power cycling is so destructive - because that one did not
first learn the facts. Failure was due to a manufacturing
defect in combination with too many hours of operation.

Don't fall for the myths about 24/7 operation. Use the
machine as designed. Power off, sleep it, or hibernate it
when done. Demonstrated by fundamental theory, experience,
contrarian logic, and real world examples is how so many fall
for 24/7 myths.

BTW, I have over there an IBM AT-PC vintage 1984.
Motherboard was upgraded but most of it, including power
supply, is original. It gets powered up often when the network
requires it functions. No failures. None. This computer is
a 1994 Compaq DX2-66 Mhz machine that is used for Internet
access. Again, never a failure. All are power cycled daily
if not more often - without any failures. BTW this Compaq is
also used during every thunderstorm. Again no failure because
I first learned why failures occur and don't worry about
lightning damage either. Its called first learning how things
work.

 

[RRR_News]

Tom,
I have two (2) desktop PC's, a '98 Packard Bell MM955 (333Mhz AMD) & '03
Dell Dim 4600 (2.4Ghz P4). Except for reboots for software installation,
hardware installation, when vacuming is done in the room, and the rare power
outages in my area. The Packard Bell has been running 24/7 since OCT'98. And
Dell for the same reasons mention before since JUL'03.

What I would consider doing, is to turn off the CRT monitor attached, when
not in use. Gone through two (2) generic made Proview ones for my PBell over
the same time period. Have no clue about flat screen monitors, but the
current ones should be better made.

Just make sure that the room is kept at a reasonable temperature, and PC is
kept dusted for proper ventilation.

--

Rich/rerat

(RRR News) <message rule>
<<Previous Text Snipped to Save Bandwidth When Appropriate>>



"Thomas G. Marshall" <[email protected]>
wrote in message
(XP SP1 / Dim 8300 / 3.0 GHz / 800 MHz FSB / 512 meg / bla bla...)

I seem to be getting a virus here and there found by NAV2003 that makes its
way in through the auto-protect. In this case I got a couple of circa 2003
Trojan.ByteVerifies. Don't know how the heck such a simple file can land on
my system, particularly since it's such a well understood virus.

I am considering leaving the system on 24/7 and establishing a daily viral
sweep.

Questions:

1. Is the 8300 cooled enough or otherwise built for staying on? I'll of
course use power options to shutdown unnecessary things and brown down
perhaps the motherboard or something. I'll have to learn more about
this---I'm half ignorant on all things power control except for hibernation.

2. Am I leaving myself open statistically to more infection simply by
staying on? I'm running SP1's firewall. SP2 is not an option currently
because of software incompatibilities.

3. Any thoughts on what I might have to worry about, in general and/or
specifically to the 8300?

Thanks!

 

[steve]

The computer left on 24/7 was finally shut off. Then it
would not power up. This computer was now considered failed
due to being 'power cycled'. But it should have remained a
prime example of why 24/7 is so destructive. In the meantime,
once we first made the distinction, then suddenly computers
left on 24/7 were failing more often.

I agree entirely. The 24/7 systems fail more often when switched on
but are only needed to be switched on once or twice a year. Ours
rarely failed during the time they were switched on. The occasional
break was usually due to a HD fails. Fans did not fail as often as I
expected they would.

We see this with computers, TVs, radios, and so many other
electronics (not to be confuses with high tech, custom
designed circuits designed to maximize other parameters - ie
Deep Space Network). If leaving computers on 24/7 was so much
better, then we must also leave on TVs and radios 24/7. One
cannot have it both ways. Either Steve must also recommend
leaving TV and radios always on, or he has misrepresented the
data by doing subjective observations.

Not at all, if it's not being used, switch it off. However, if it will
need to be switched back on a few minutes later, leave it on!

Bottom line: once we confronted how these people were
'collecting' the data, then suddenly numerous computers listed
as 'failed due to power cycling' were transferred to the
column that said 'failed due to 24/7 wear'.

The development systems I mentioned were left on all the time because
they were in use all the time running real development work round the
clock. I was one of the people using them. The admin machines were
switched off when they were not required. That was every night, all
night. It's not a random sample of systems. These are two sets of
machines. The admin systems are in the office environment, with almost
no regular maintenance. The development systems are in large computer
rooms and share the mainframe environment to some extent. They are
also kept clean, that may explain some of the differences.

The data was collected from the logging system kept by the maintenance
teams that serviced our computers. I wrote the logging system about
twenty years ago.

 

[User N]

Julian coughed up:

No, seriously, the reason for leaving it on 24/7 is that I *don't* want the
AV scan to run while I'm actually using it. []

You seem to want a "AV" scans to be performed every day. Specifically,
what "AV" software are you referring to? Is it something that lacks the
ability to automatically scan on an as needed basis? For example, when
the filesystem opens [or closes] files?

 

[Paul Schilter]

Thomas,
If your knowledgeable about hibernation then you're ahead of Microsoft.
Paul


Thomas G. Marshall wrote:
snipped

---I'm half ignorant on all things power control except for hibernation.

snipped

 

[Paul Schilter]

w_tom,
If the fan failed due to bearings that would be very easy to observe
when you remove it. If it spins freely and smoothly, the bearings are
fine and the motor failed. If it doesn't rotate well then the bearings
failed and could have then caused the motor to fail.
Paul

 

[w_tom]

So you recommend running the system and consuming all that
power to only get the same reliability as when a system is
powered down at the end of the day?

24/7 operation provides nothing significant to system
reliability, causes increased component wear, and consumes
electricity to no purpose. Wear from hours of operation is
significant on the parts that fail most often. Power up does
not cause wear despite the many myths to the contrary. Those
who promote power up as destructive do not tell us which
electrical parts failed or why they are damaged. IOW a powerup
surge is mostly wild speculation. Why consume electricity to
no purpose?

A computer should work just fine in a 100 degree room as it
does in a 70 degree room. In fact, I don't use air
conditioning. Hate the stuff. These computers of 10+ years
in constant use get run even when temperature is 100 degrees.
They still don't fail. Same is true of dust problems. Dust
problems are greatest when the naive start adding fans to
solve the mythical heat problem. Too many fans don't cause
any appreciable cooling but can cause excessive internal dust
balls. Problems created by heat and dust are common myths.

Run a new computer one day in a 100 degree room. If the
computer has marginal or intermittent components, those
defective components are best identified before the warranty
expires. A 100 degree room is not destructive to a computer
despite so many myths to the contrary. Heat is an excellent
diagnostic to find defectives before hardware fails obviously.

I've been doing computers and aerospace electronics for too
many decades to fall for these well promoted myths about 24/7
operation, heat, and dust created problems. Too many years
and too much asking questions at the electronic component
level.

 

[w_tom]

Fan failure and bearing wear, as described in the earlier
post, could not be easily identified by restricted spin.
Let's get down to component level analysis. That Hall Effect
sensor may not be sitting flat on a PC board inside that fan.
Torque is significantly reduced. With but slight bearing
wear, now that fan would not have sufficient torque to startup
every time. Using a soldering iron, I once reseated a Hall
effect sensor so that it was not tilted more than 15 degrees.
A flat sitting sensor increased fan torque significantly so
that it could overcome an undetectable amount of bearing wear.

Just because a bearing still feels free and smooth does not
mean the hall effect sensor is properly seated or that a fan
has not suffered some bearing wear. Only slight bearing wear
could make the fan intermittent (another example of why QC
inspection does not create component and therefore system
reliability). Fan failed due to a combination of too many
hours of operation AND a manufacturing defect. Others may
erroneously assume fan failed due to power on surge.

BTW, fan tends to be a more frequent failing component. So
we install two fans working in series. One blows in. The
other blows out. Then when one fan fails, the other will
maintain airflow until a human finally discovers the problem.
One fan provides sufficient cooling. Two may be installed
only because fans fail so often with mechanical wear.

Yes a fan with massively worn sleeve bearings can be
apparent. But that is not the only reason why bearing can
cause fan failure. Notice the point repeatedly made. One
does not know why failures occur only because they observe.
Observation can report when a failure has happened. One must
also understand reasons why that observed failure happens to
learn something useful. The devil is in those technical
details. Many who recommend 24/7 operation for reliable
operation don't first learn underlying theory. Therefore they
can be deceived by what they have observed.

Another way to improve reliability (which is why those
Hondas and Toyotas also had higher reliability)? Keep those
human hands out. Humans are another major source of failure
when we demand the reliability I call minimally acceptable
even for household appliances. We once wanted to clean out
the massive dust inside the software development system (in my
naive and younger days). Manager of software development
would not let us - and for good reason. Major dust balls were
not a problem. But human hands do create new and intermittent
failures. Cleaning can even create problems despite a human
emotion that a clean computer is better. Don't let human
emotions or human hands reduce system reliability.

One more interesting fact. If a vacuum cleaner causes
computer failure, then the computer has internal hardware
problems. Again, too many will blame the vacuum rather than
first learn how (or why) failure happens. Therefore one may
blame the vacuum rather than a defect inside the computer.
Just another example of how observation alone leads to
erroneous conclusions. Just another example of why we want
everyone to study junior high school science. To know
something, one must first learn both the underlying theory and
obtain supporting experimental evidence. Anything less is
called junk science reasoning or wild speculation.

 

[Thomas G. Marshall]

[top posting fixed into bottom posting, so I could more sensibly snip]

w_tom coughed up:

The computer left on 24/7 was finally shut off. Then it
would not power up. This computer was now considered failed
due to being 'power cycled'. But it should have remained a
prime example of why 24/7 is so destructive. In the meantime,
once we first made the distinction, then suddenly computers
left on 24/7 were failing more often.

We see this with computers, TVs, radios, and so many other
electronics (not to be confuses with high tech, custom
designed circuits designed to maximize other parameters - ie
Deep Space Network). If leaving computers on 24/7 was so much
better, then we must also leave on TVs and radios 24/7. One
cannot have it both ways. Either Steve must also recommend
leaving TV and radios always on, or he has misrepresented the
data by doing subjective observations.

Bottom line: once we confronted how these people were
'collecting' the data, then suddenly numerous computers listed
as 'failed due to power cycling' were transferred to the
column that said 'failed due to 24/7 wear'.

Yes, I followed this reasoning already from your prior posts.

First, Steve was not talking about a sampling of one. He was talking about
several computers in two distinct groups: turned off, and 24/7.

Second, he reports that over a *period of time* he saw the turned off
machines fail much more often than the 24/7 ones.

Your logic addresses what category someone might place a machine at the
precise moment a failure is discovered. That is not what Steve is doing at
all.

I still don't see you address the bottom line issue: The ones that were
left on 24/7 failed far less often. Yes, it depends upon what category the
failed machines are placed in. But when you *start* with the category and
*then* count the failures for each, as steve did, he ends up with the
following:

For any given time span of your choice:
24/7 machines: X repairs
on/off machines: much more than X repairs

....which I believe is a fair assessment of his observations. I still don't
see you counter this directly. I can only assume that you believe that
steve has made a fundamental error here.

But it's /not/ this error:

Steve sees a machine fail at power on
Steve then assumes it's because of power on

It is /this observation/ (without error):

Steve sees two groups of machines: on/off and 24/7
Steve sees the 24/7 with far fewer failures.

....[rip]...

 

[Thomas G. Marshall]

User N coughed up:

"Thomas G. Marshall"

Julian coughed up:

No, seriously, the reason for leaving it on 24/7 is that I *don't*
want the AV scan to run while I'm actually using it. []

You seem to want a "AV" scans to be performed every day.
Specifically,
what "AV" software are you referring to? Is it something that lacks
the
ability to automatically scan on an as needed basis? For example,
when
the filesystem opens [or closes] files?

As I said in the original post, it is NAV2003.

 

[Thomas G. Marshall]

lol....


Paul Schilter coughed up:

Thomas,
If your knowledgeable about hibernation then you're ahead of
Microsoft. Paul


Thomas G. Marshall wrote:
snipped

---I'm half ignorant on all things power control except for
hibernation.

snipped