|
Window Server R 2000
We have several servers with a singular purpose of running a display application. No one depends on them but the developers and the analysts during mission time. I like to turn them off when not in use and over the weekend. Others say servers are designed to be run 24/7. That statement can be re-worded to: Servers are not designed to be turned off.
I do not agree with that.
I suspect that all computers are designed to run 24/7. Would you design one that can only be run 8 hours before being turned off? So how might a server, or any computer, be designed differently to run 24/7? Or designed to not be turned off?
By turning it off we save power running the server, we reduce the total amount of dust and air borne debris moved through the server by the fans, we reduce the run time drives by 2/3 or more, and we save the money that would be spend removing the heat from the room.
On the negative side, you gotta push the button and do something else for a few minutes before it is ready.
I have had PCs since about 1983, turned them off every night, and never had a failure, drive or component.
So, Please, what is your opinion? Please reference any supporting information.
Thank you for your time
|
|
|
|
|
The hardest period of time during a computer's life is when it powers up and when it powers down. This is especially true for disc drives that either move the heads to the landing area and spin down or spin up and move the heads on to the data surface. Then there are the surges of power into the PSU and then the motherboard and the rest of the components. In my experience, I have never had a component fail during normal operation. The only failures I can recall are during power on sequences. For this reason, I try to minimize them.
|
|
|
|
|
It's uncommon for a light bulb to blow while it's still on, they usually fail when they're first turned on.
But you wouldn't leave a bulb on all the time to make it last longer (there are of course, exceptions[^])
What you'd need to know is where the crossover point is, how long does the computer need to be idling to match the stress of it starting up?
And if you are able to prolong the life of your computer by leaving it running, does the cost of keeping it powered on outweigh the cost of replacing a component if it does fail?
|
|
|
|
|
Anthony Mushrow wrote: It's uncommon for a light bulb to blow while it's still on, they usually fail when they're first turned on.
But you wouldn't leave a bulb on all the time to make it last longer (there are of course, exceptions[^])
Only that computers are slightly more complicated beasts than light bulbs. When a power supply fails, it can easily fry the entire computer. Therefore you need a good power supply that has reasonable protection against this and then let it do its job. Who cares when it fails after some years? It's easily replaced as long as it does not take the whole computer with it.
Servers often even have a dual power supply to keep going when one fails. You can replace the defective power supply (or even drives or memory) while the server is running. I really don't think all this is a real problem. Servers may be designed to keep going, no matter what happens. That does not mean that they will instantly fall apart when you shut them down when you don't need them.
Dang, I have a computer that almost 40 years old and still runs with its old power supply and its old monitor. I don't know how often it has been torn apart (sometimes by people who don't know that CMOS is not an exotic dish), put back together again (sometimes after years in a box) or been switched on or off. Once I even had accidentally reversed the polarity of the power supply, giving the power supply a mighty start and the voltage regulator (a good old LM7805) got a little hot.
The language is JavaScript. that of Mordor, which I will not utter here
This is Javascript. If you put big wheels and a racing stripe on a golf cart, it's still a f***ing golf cart.
"I don't know, extraterrestrial?"
"You mean like from space?"
"No, from Canada."
If software development were a circus, we would all be the clowns.
|
|
|
|
|
The bulb was just a simplified example. I'm just challenging the idea that leaving a computer on will save it some stress and make it last longer. There will be some duration where turning off the computer will be better for it than leaving it on and idling, but how long would that be? An hour, a day, two weeks? I have no idea.
|
|
|
|
|
The items I was referring to that are stressed the most during start up and shut down are the mechanical items - disc drives and fans for example. Electronics also get stressed from the inrush current but not to the degree that the mechanical parts do. The stresses during these times are much more severe than the steady state stresses and take a much bigger chunk out of the lifetime of a product. At least, that is what we were taught at my engineering school.
|
|
|
|
|
For new hardware, its probably not an issue, but if its old hardware then it is best to keep them on.
The disks prefer to be kept spinning and the electronics prefer a constant temperature. The cooling and warming can slowly start to cause issues.
modified 26-Apr-17 23:45pm.
|
|
|
|
|
I'd say this also comes from applications running on servers that need to have high availability and are even working overnight. Just take a database server as an example.
But yeah, they are design to be able to work 24/7 but you surely can turn them off if not needed, though as already mentioned before, the older the hardware gets the less it should be powered down.
For example we had some switches 15yo and they ran perfectly, sadly after electric breakdown they werent able to start up again. Hardware died.
Rules for the FOSW ![ ^]
if(this.signature != "")
{
MessageBox.Show("This is my signature: " + Environment.NewLine + signature);
}
else
{
MessageBox.Show("404-Signature not found");
}
|
|
|
|
|
When working with electronics usually it is a good idea to keep them ON.
The power cycle is the most demanding for that kind of devices.
If you keep those servers without display they are not consuming a lot of power, and you could always virtualize them...
|
|
|
|
|
Hi,
Keeping servers running 24/7 does prolong their lifetime. The motherboard and peripherals hardly gain any benefit at all... only the capacitors are susceptible.
However... the hard drive is s different story.
Failure Trends in a Large Disk Drive Population[^]
TL;DR: Hard drives fail faster if they are power cycled.
Many years ago I made a Russian system administrator cry when I rebooted a Linux box that had been up for 4+ years. No, I am not kidding.
Best Wishes,
-David Delaune
|
|
|
|
|
Randor wrote: Keeping servers running 24/7 does prolong their lifetime. The motherboard and peripherals hardly gain any benefit at all.
When the machine is off there is a chance insects (ants, spiders) may move in,
- when machine is on it's too hot & dry for them.
The support chips on the mobo have such thin legs ants will actually bite through them,
and spiders are actually quite wet (conductive).
Very small chance, but I've seen both happen.
Sin tack
the any key okay
|
|
|
|
|
Thanks for the link. I did notice this in the study:
As is common in server-class deployments, the disks
were powered on, spinning, and generally in service for
essentially all of their recorded life.
That makes this study not applicable to the fundamental question.
Thank you for your time
|
|
|
|
|
Power on hours versus on/off cycles has been argued for years, especially when it comes to motors (fans, disk drives, etc).
The most important thing to consider is never letting the smoke out. Electronic devicess are powered by smoke. If you ever see the smoke escape, they no longer work. 
Arguing with a woman is like reading the Software License Agreement. In the end, you ignore everything and click "I agree".
Anonymous
|
|
|
|
|
If they're vintage hardware from 15+ years ago IMO it's a minor miracle that they're still working as is. If it's just apathy that they're still running on museum pieces I'd strongly recommend migrating to new boxes running a modern OS or virtualizing the apps to do the same. 15YO hardware is on borrowed time no matter what you do; new hardware won't care either way and VMs are like: "LOL WUT Hardware?!?"
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, waging all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
My work PC is never turned off because I need to be able to access it without waiting for someone to have to manually power it on. The monitors turn off automatically soon after I lock my workstation for the day. I always shutdown my home PC at the end of the day. Have never experienced a hardware failure. Yet. 
/ravi
|
|
|
|
|
There are valid points made here for both sides, so I'll skip that. I just wanted to let you know that if you DO shutdown stuff, I have written a pretty nice application for doing remote WOL and shutdown. It supports waking up remote servers across the Internet (I do that myself daily). I offer it free and open source to everyone. https://wol.aquilatech.com/
|
|
|
|
|
Well thats cool. Have not looked at it but: Thank you.
Thank you for your time
|
|
|
|
|
Lucky you.
I have lost hard drives, Graphics Adapters, Memory Chips (causing BSOD while running).
But the scariest thing I have ever done was cycle a few Novell servers that had been running for years.
Out of about 10 of these old DOS based network servers, 2 never rebooted because of hard drive failures.
Most had errors in their startup files, from changes that were NEVER tested. And MANY had never been properly backed up.
Trying to explain to a business owner the risk of turning off his Server, and explaining that the hardware is rated for 5 years, he has had it for 8 years, and has not rebooted it in 3 years... Your luck might be ready to run out.
I, too, used to turn my big machine off every day. But I now leave me machine running all the time. My monitors are turned off. The machine self-regulates its power consumption. But a BIG part of turning the machine off before was Power+Noise+(The OS would CRASH if left running too long, like 25hrs, LOL)
|
|
|
|
|
Today, it's "security robots"[^]. Tomorrow, it's ED-209 from Robocop saying "You have 20 seconds to comply!"
On the other hand, you have different fingers. - Steven Wright
|
|
|
|
|
Hmmmmm,
That thing has been rolling around Microsoft campus since 2014[^]. It's usually creeping around real slow... but it can move fast too...
Best Wishes,
-David Delaune
|
|
|
|
|
Wow! I had no idea. So where do they hide the guided missiles in that thing? 
On the other hand, you have different fingers. - Steven Wright
|
|
|
|
|
That's why we're keeping staircases.. 
The Fall of ED209 - YouTube
Now is it bad enough that you let somebody else kick your butts without you trying to do it to each other? Now if we're all talking about the same man, and I think we are... it appears he's got a rather growing collection of our bikes.
modified 31-Aug-21 21:01pm.
|
|
|
|
|
The best is seeing robocop going down afterwards...
I can't still decide which of both robots is going downstairs more elegantly...
|
|
|
|
|
Slinky-robot.. it's the way to go
How to make a slinky go up the stairs. - YouTube
Now is it bad enough that you let somebody else kick your butts without you trying to do it to each other? Now if we're all talking about the same man, and I think we are... it appears he's got a rather growing collection of our bikes.
modified 31-Aug-21 21:01pm.
|
|
|
|
|
...Protect the innocent ... Uphold the law ...
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
