Desktop·25,135 discussions

Does restarting desktop each day have any adverse impact?

Locked

For a long long time, I have the habit of restarting my machine before I leave work, at the end of the day. Before this habit I used to shut the desktop off.

My coworker said that sometimes “restarting” everyday could adversely affect the computer. He talked about variances in temperature and what not. To put it simply I did not understand 🙁

So here is my question: Does restarting the desktop each day reduce its life compared to restarting it less often? Or does it in anyway have a negative impact on the desktop?

Topic

I have all

In reply to Does restarting desktop each day have any adverse impact?

of our users Restart every day. Reasons are, even with XP, the programs will run better with less sluggishness by restarting each day. In what you friend says is true if you do a Shut down. I do agree with that. However, every time one of the docs calls me and says one of the PCs in the patient rooms is running real slow I go to check it and it has not been logged off for several days. I restart it and, wow, just like magic, it works great. I also have several programs that update and run every night so they have to be on. In my book the computers, at least in the work place, should be restarted each day.

Edited to add: we keep our computers until they die a terrible death. We have had these with XP for three years and have not seen any adverse effect of Restarting.

Sluggishness ?

In reply to I have all

Is there any sluggishness due to just hardware issues?

When I have checked the

In reply to Sluggishness ?

Memory usage after the PC has been logged for several days it is almost maxed with the amount of RAM. Restarting the workstation dumps everything out and helps it to run better. Even closing out the programs doesn’t clear the memory usuage all the way.

But…

In reply to When I have checked the

…this is software related, right?

In my mind,

In reply to But…

which has a lot of room to get lost, the software is using the RAM for working. Sometimes the software doesn’t turn loose of its portion of the RAM even when you close it out. What I have experienced is that doing the restart each day the users have a lot less issues with sluggishness.
Edited to add: I think it is both but one could argue it’s software. Just a thought.

A question for you

In reply to In my mind,

Howdy…

A question for you…

[i]”Sometimes the software doesn’t turn loose of its portion of the RAM even when you close it out.”[/i]

Are you saying that you have observed something that indicates when a Windows XP process (program) is shut down, that its memory is not released? Can you describe how you determined this?

In Windows XP, each program runs in its own virtual memory address space. Shutting down the program removes the address space and the memory pages (paged in or out) are made available for reuse.

Now, there are persistent common SYSTEM memory areas that running applications can cause to become enlarged. These persist beyond the life of an individual address space. If you’re experiencing problems with this, more physical RAM should completely solve it.

My experience: With Windows XP workstations running 1.5 or 2.0 GB of RAM, they can stay up indefinitely without ever slowing down — weeks or months.

Thanks in advance.

edit: grammar

My Answer for you, Yes.

In reply to A question for you

Yes, I have seen where you can close out a program and it does not release all it’s memory allocation. This article is regarding User Hive issues but in the overview it states the same http://tinyurl.com/2h3yl .
All of our workstations in the patient areas are running 2GB of RAM. The longer (days) they stay open the slower it gets. Just closing the programs will release some of the RAM but a complete restart will let it start off, usually >250MB. It could be a direct correlation to the software but I couldn’t show that. My statement is it is better, at least for our systems, to restart each day. Even the users can tell a dramatic difference.

BTW, TechExec2, on a personal note, I have noticed a tremendous change from the time you first started posting in TR and now. Good change. 🙂

Techexec2

In reply to A question for you

Windows is supposed to work as you say, but often it does not. I have 3 work systems, and for the most part they do not need to be rebooted during the work week. However, sometimes they do.
Windows does not release the memory address if a thread of the program is still running, especially if the program called a file (like a dll) and there was no proper close to it. This memory is ‘in use’ even when nothing is actually running or using it.
Also, Windows does not always release temp files properly either, nor does it release page file memory properly (often). Unless using 3rd party tools, or altering some registry entries. there are many places that Windows fails to free up memory and/or clean the page file.
Shutting down in the eve is a good practice, leaving them on as I do is not. But with the network problems recently, I fear that I will not be able to log back in if I shut down (happens often) for a few hours. And I hate being down completely.

Old Guy: Memory leaks

In reply to A question for you

Maybe the software applications are leaking memory. It is not uncommon for the code to release its memory cleanly. If it is not the software application itself, then it could be an issue at framework/platform/OS level.

OldGuy: Thanks for your reply

In reply to A question for you

My read of the MS “User Profile Hive” article is that it is only talking about incomplete termination of user sessions when a user logs off and not being able to log in. The article doesn’t mention memory allocation problems, although one could assume some memory was not freed also.

W2KTECHMAN’S POST

In w2ktechman’s post (link below), he mentions things that are about system memory allocations (memory shared between processes) that are sometimes not properly released when an application (process) is shut down, and some other problems.

So, this still looks like system memory allocation issues (and Windows bugs in handling them), not the process’ virtual memory address space per-se. And, a good way to resolve that is to reboot.

But, it still leaves unexplained how I can get away with rebooting only after weeks or months, and never to resolve slow-downs or instability (always for some other reason like a software install or testing an aberrant pre-release program).

ONBLISS’ POST

I think onbliss (link below) is right (VERY presumptive of me since it is YOUR shop! :0 ). It is likely your applications are leaking memory in their own address space. And, merely restarting them should produce an great result. But, having done all that, going one step further and rebooting too makes a LOT of good sense.

REFERENCES

w2ktechman’s post
http://techrepublic.com.com/5208-11183-0.html?forumID=5&threadID=200958&messageID=2094720

onbliss’ post
http://techrepublic.com.com/5208-11183-0.html?forumID=5&threadID=200958&messageID=2094740

edit: spelling

w2ktechman: Thanks for your reply

In reply to A question for you

…

Old Guy: Thanks also…

In reply to A question for you

[i]”BTW, TechExec2, on a personal note, I have noticed a tremendous change from the time you first started posting in TR and now. Good change.”[/i]

Thanks also for your kind words. I’m not completely sure which specific posts you are referring to (but I can guess! 🙂 ). Like events in the “meat world” (LOVE that term), there are more memorable ones and some you would prefer to forget. And, most of us learn and grow as we go. TR [i]is[/i] a different kind of place on the web that has its own rhythm. And, observation of anyone here is done as if through a straw — It’s difficult to get a wider view.

Good Questions

In reply to A question for you

1 way to limit the memory problems is to use only a few applications at a time. Many applications share Windows files, or Dll’s in the library. When 2 or more of these programs are sharing a file, it may not close properly when 1 or all of the programs stop using it.

Also, often a program does not release some threads that it starts, and they may be harder to find. This is due to bad coding, but since I am not a programmer, I will not comment further.

Other portions of problems are with Windows itself. The default setup is meant for an average person, who will shut down their system after its use. Up-time is usually only an issue with server and work environments. There are several registry enhancing tools out there, and they do work. But most of them also use a generic explanation to make it sound like it is always better to choose 1 option over another. This is a bad approach, they should explain the differences better (I havent used one in a few years, this may have changed).

Keeping your startup applications to a minimum will help as well. So will periodically defragmenting your system, cleaning temp files, and cleaning the prefetch folder. If you do not know much about the prefetch folder (c:\Windows\Prefetch), whenever you open an application, a small portion of it copied to the prefetch folder. From then on, every time you boot, it will load a portion of each program in there (why is beyond me that this is default). So over time you can have hundreds of partial applications running in memory, for things that you have only run or used even 1 time. I clean out this folder about once a month, sometimes a little longer.

Also slowdowns may be because of more malicious software installed. Check for buggy programs and or spyware regularly.

If you notice that certain program combinations lead to the instability, then do not use these programs at the same time. Windows XP does not need to be rebooted often. The exception is the introduction of other programs, and combinations of programs running at the same time.

I hope this helps out.
By the way, I have used this program to help out. http://tweaknow.com/products.html
look for Ram Idle

Memory release

In reply to A question for you

That’s code quality issue particularly with 3rd party apps.
I can’t remember the last time I traced a leak in windows to something internal. For all their faults MS have made a good fist at resolving those issues in terms of the base OS.

Leaky apps are a whole different ball game though. Memory lost within the process will come back if you restart it. The no choice but to log off / reboot leak, comes from out of process allocations.

A classic is to raise a thread to show a starting animation and not release the handle. While that’s still alocated windows keeps the memory, close the app and now the record of the handle that needed releasing gets released giving you all four bytes back. The thread and all it’s memory of course is still allocated.
That can get quite expensive after a while and exacerbates the issue with other leaks. I fixed one applicaton that leaked 500k out of process every time you ran it, and all the memory to hold the data is was working on every time you loaded a new entity.

It was impressive in it’s own way.

The fault I was presented with was an intermittent av when a particular order was loaded which took a long time , that order being the biggest one in the database.

Needless to say, I was looking for the fault in the wrong place for a fair while as the top technology firm I was working for would never implement such a juvenile piece of crap.

What?

In reply to A question for you

Any knowledgeable IT person would know that Windows itself, and many apps, don’t release all of the memory from time to time and that processes run amok leaving memory in “no man’s land”. I have seen with ALL Windows versions, even XP, where a reboot solves over 50% of all problems, a lot of the memory issues from being left on for a few days. I have yet to see a computer running XP go more than two weeks without at least a 20% degradation is performance.

Getting back to the original question: restarting every day has no big impact on a machine. I know of machines that lasted over four years doing that with no issues. I will say that leaving a machine on at night, unless absolutely necessary, is a major security issue.

MD: You’re very impressive

In reply to A question for you

Any knowledgeable IT person would know that experiences vary widely depending upon all of the many variables, especially when it comes to Windows. Got extensive empirical data behind those precise percentages, do you?

My advice: Don’t be too impressed with yourself. Such overconfidence is not fooling anyone. Most of us learned that before we left our 20’s.

Reclaim Your RAM – Fast and Easy

In reply to A question for you

Regarding shut down vs restarting:

I worked for a large company and their studies found it to be substantially more economical to shut down pc’s at night to save the energy costs. Leaving them on all the time/overnight was a substantial cost factor.

Restarting the pc is definitely a good way to “refresh” the pc and reclaim memory and is a very good option. The restart doesn’t stop the drives from spinning etc. so there is no startup shock to the hardware components at all.

Here’s a little method I use on my home pc to monitor and reclaim RAM.

Firstly, it’s a good idea to know how much RAM is on your PC. Go to http://www.belarc.com and download and run their free Belarc Advisor software. This will give you a really good summary report about what hardware and software you have on your PC.

Go to http://www.download.com and search for FreeRAM XP Pro and Instant Memory Cleaner. Download and install them both.

FreeRAM XP Pro will put a digital RAM meter in the system tray. Instant Memory Cleaner will put an icon there also.

Right click the FreeRAM meter. Click on RESTORE and then SETTINGS and then on the DISPLAY tab and check the CPU Usage box to display that also when you open the FreeRAM window. This meter is a great way to be able to monitor your RAM availability in real time.
If for some reason you want to uninstall the RAM meter; right click on the meter and then click on INFORMATION at the top of the window, and then on Uninstall. This program will not show up in the Add/Remove Programs of Windows.

If, after some hours of using your PC, you notice that the RAM is getting low, click on the Instant Memory Cleaner icon and run the program. It will take about a minute to run but the amount of RAM reclaimed and made available will impress you.

Here’s a quick example. I run a program called Spyware Doctor. At bootup it takes about 50M of RAM. If I run its’ scan program it jumps up to about 70M of RAM. After I run the Instant Memory Cleaner it drops to about 7M of RAM.

You can do a test yourself by first looking at what each .exe file is requiring for RAM by viewing the Task Manager (Ctrl + Alt + Del) and clicking on the Processes tab and review the Mem Usage column. Check the Mem Usage immediately after bootup. Write down the .exe file names and the amount of RAM they are using. Then run the Instant Memory Cleaner and review the same .exe files and see how much RAM is being used by each .exe afterwards.

Hope this helps.

Memory Usage

In reply to A question for you

Utilities like Instant Memory Cleaner do nothing but force pages out of RAM into the paging file and trims working sets (which actually may or may not be a good thing to do depending on the particular process) – it just runs MS’s ClearMem utility in the background.

It’s not really freeing up any “wasted”/”lost” memory as suggested earlier.

The OS’s virtual memory subsystem allows processes to retain used pages of their virtual space in RAM to the extent of their working sets and to the extent of other system RAM usage conditions/needs. Also note that if all you’re doing is looking at the “available” memory statistic, you’re not getting a clear picture of memory – the system automatically maintains a cache of memory pages. These are recycled if needed but act like a level 2 cache outside of processes so that if a process needs to page in a page that’s in the cache the system will give it to it directly without incurring a disk drive page from the page file (or a read-only page from an executable image). These cached pages are also automatically recycled if needed for other things on a least recently used basis. They are therefore really part of “available” memory.

Terminating a process kills all of its threads AND deallocates all memory allocated to it – it’s part of the kernel’s process rundown routine. It’s not possible for a terminated process to leave stranded memory in no-man’s zone as someone suggested.

A user process may “leak” memory, but only to the extent that it will consume more and more of its virtual space and any physical memory allocated to its active working set – however this is deallocated when the process terminates. A user process cannot “leak” memory permenantly beyond its lifetime.

However, 3rd party device drivers/system-level code (printer, network, video, special purpose, etc.) can and often do leak memory permanently – some video and network drivers can actually leak fairly large amounts of (non-pagable) memory. Drivers don’t undergo process rundown and can therefore permanently leak memory if poorly constructed (until a system restart).

If your system slows down with use, it’s almost certainly related to lousy 3rd party software – not an issue with XP.

I go months without shutting down or even logging off my PCs (I just lock the workstation when I’m not there) with no ill effects, slow down, loss of RAM, etc. – and I use everything from the usual MS Office and IE to Visual Studio to image processing to video editing.

I agree with Old Guy

In reply to A question for you

150%!! My challenge? Explaining the premise to my users in non-technical terms!

All the current released versions of Windows

In reply to A question for you

don’t manage memory well. The operating system does not automatically clear out the memory of closed programs, Windows relies on the application to send the commands to do that.

When I went to tech college to get some degrees in the mid/late 1990s, a teacher demonstrated this on the college PCs. We stood around while they opened a PC, then opened Task Manager, switched to the process screen. We watched this as he opened MS Word, closed it, opened it, closed it, repeating this several times. Each time it was opened some new processes were started and only some were left open after it was closed, a new instance of them was opened when he opened Word again. Also some processes kept increasing in the amount of RAM and VM used, but did shrink at all. After opening Word nine times the system had a BSOD.

Now that was with a Pentium running Win NT 4. But you can do the same thing with Win XP, or better yet, install a memory mangement program and watch the amount of RAM keep dropping.

The worst applications for this problem were those made by MS themselves. For an interesting experiment, restart your system and don’t start any applications until after you have Task Manager running, with the processes showing. Then open applications, noting the processes they start, and which close with them.

MS have a process called Messenger (not the application MS Messenger) that can be turned off in the services. Disable this, Outlook Express and MS Internet Explorer will still work fine. Enable it, at every time you open OE or MSIE, it will open and stay open after you close OE or MSIE. And that is just one clear example of this problem.

BTW 2rs – I explain this problem to non-techs by having them imagine the RAM is like an old style mail sorting desk with all the little boxes for mail, when an application opens it loads information into the available memory boxes for use; when it closes, a good application knows which boxes it used and clears them out, while a poor application doesn’t do this well and leaves some info in some boxes. After this happens anough times there are not enough empty boxes available for a new application to use when opening properly, so it doesn’t open properly and a fault is generated. A system restart is like someone going a long and clearing out all the boxes to the rubbish bin.

Here’s a test for you

In reply to A question for you

Run NetBeans 5.0, code up a simple command line program and execute it within the IDE. Rinse and repeat 5 times. You’ll begin to notice quite a bit of sluggishness, even if you close the application.

This happens to me frequently, and I’m on XP with 1 GB Ram. Usually, I also have other applications open (the usual memory hogs — TOAD, Visual Studio, Remedy, Outlook), so you might want to mimick this scenario.

Most of the time it is badly written apps…

In reply to In my mind,

…from MS and other companies that write to RAM and that don’t release properly when the app is closed. Memory leakage has been a problem with Windows from the very beginning and I don’t foresee it being cured with Vista although MS’s memory management tools have improved with each new generation of Windows.

As far as restarting the computer every day goes… no problem whatsoever with any of the 70 systems in my company that are my responsibility. I have finally been able to convince my users (because they see the proof themselves) when I tell them that a cold boot will flush the RAM and give them back the speed they had. To some of them, it seems like witchcraft. But hey… if it works… what the heck.

Restarting after some time

In reply to I have all

Like a normal human being a computer needs to rest so restarting it once in a while does help the efficiencing of the normal processing because you are clearing the memory by doing so and reducing the number of processes that are waiting to be completed.
If you keep your computer on for a very long time without switching it off for atleats 8 hours you may endup frying the cooling fan and therefor you processor would also become damaged so it would be wise to switch it off every 3 days and give it a break for atleast 8 hours

It’s only MS Windows that needs to be handled like this

In reply to Restarting after some time

other operating systems allow you to restart individual services and processes without disturbing the rest of the system. I’ve been using micro and personal computers for over 20 years, and they do NOT need to be turned off and allowed to cool down, UNLESS you have poor quality fans, the good ones are designed for many years of 24/7 running, and even the bad ones run better left on all the time.

What is important, is to make sure the airflow is sufficent. Putting humungeous fans it doesn’t help if you then stick the PC in a cupboards with insufficient air flow. Good fans and plenty of space around the box, and they last for years.

I know of old 386 servers that have only been off twice – major power failures and the back up batteries ran flat. Last year they got upgraded to P4s the owners don’t expect to turn them off until the next upgrade, say 2025 or so. They use long term planning.

My main machine aP4 3ghz is on 24/7 and only gets turned off when we lose power, or I go away for the weekend. That’s a total of 7 days off in the last 3 years. Being Linux loading new software doesn’t require a full restart.

I wondered who would start the MS slating……………

In reply to It’s only MS Windows that needs to be handled like this

Where you are talking about hardware you would be correct if you said x86 platforms needed to be treated like this in the old days. Modern machines do not need to be turned on and off due to temperature effect unless as you correctly state, there is insufficient cooling in and around the computer.

Microsoft is blameless for this………..

One of the big problems that occurs in Windows

In reply to I wondered who would start the MS slating……………

is that the OS does not properly manage, and clear, the RAM when a program closes down. Many third party applications include a routine to ensure this happens, but very few, if any, MS applications do. Watch how much run a program to check on your available RAM and Virtual Memory, then open and close MS Word, Excel etc – the amount of used VM and RAM will keep growing, not drop back when the application closes. You have to periodically reboot the system to clear this out, unless you’ve installed a third party applications to manage the memory and do this for you.

So should’nt we

In reply to One of the big problems that occurs in Windows

…blame the applications such as MS Word, Excel etc that do not do a good clean up?

Yes you could, except memory management is usually an OS function

In reply to So should’nt we

and handled by the OS in most other Operating Systems, I say most as I haven’t tried every other OS – but those I have tried do have a function that monitors what memory is used by what application and clears the memory when the application closes.

But regardless of blaming the application or the OS, MS writes both and they intend their applications to only be used on Windows, so they should have such a routine in one or the other – yet they have it in neither. Thus, QED, MS is to blame either way.

edited to fix typo that changed the meaning of the sentence – amazing what a missing ‘e’ can do.

You shouldn’t have to restart

In reply to I have all

I’ll agree — these are Windows machines, but you shouldn’t have to restart them daily to “reduce sluggishness”. The real problem is that whatever software you’re using has memory leaks.

When I leave my corporate laptop at work, it remains docked and powered on. I don’t ever experience sluggishness unless I have a ton of applications open at one time and those applications are memory hogs (TOAD, Remedy HQ, various browsers, Outlook, Visual Studio, NetBeans 5.0, a few Query Analyzers, which aren’t so bad, but combined with the others…)

Usually if I get rid of TOAD and Remedy, things return to normal. Very rarely am I in the position where I have to reboot (usually after a bout with NetBeans, I’ll be forced to reboot). Being a software developer, I tend to like keeping my active code and applications up so I can pick up where I left off.

When I take my laptop home, or take my personal laptop anywhere, I simply hibernate the machine. When I bring it back up, my applications are still open and my code is still in place.

Rebooting was something for Windows 98 and 2000, but I’m fairly happy with XP because I haven’t had to constantly reboot. When I do, 98% of the time, it’s due to some software package not returning resources effectively.

You could be right in it being crappy apps, however

In reply to You shouldn’t have to restart

the majority of situations where I’ve seen people need to restart to get rid of sluggishness, is when they’ve been using MS Word, MS Excel, and the rest of MS Office. It’s not so bad if you open them, and leave them open all day, but open and close when finished, it has a real issue in XP, just as it did in all the previous versions of Windows.

Office 2003?

In reply to You could be right in it being crappy apps, however

I can’t speak for the older Office products because I really can’t remember, but I currently run Office 2003. There are several times throughout the day that I open Word, do my thing, and then close it and haven’t noticed any problems. For what it’s worth…

More true with Hibernation

In reply to I have all

I have found that even though some users hibernate their Windows XP machines that the amount of gash that gathers in temporary files and so on slows the machine down drmatically. I would disagree though that temperature differential damages PC’s. True on the old machines but build quality today means it has less effect.

Theory vs. reality

In reply to Does restarting desktop each day have any adverse impact?

After supporting computers, servers and desktops, for over 20 years I can say that very few desktop computers die prematurely due to being power cycled once a day. Save electricity and turn your desktop computer off at the end of the day.

Question,

In reply to Theory vs. reality

When do you have the computer update your software, ie., anti-virus, Windows, etc. Also, I have often read that keeping a constant temp and avoiding the little power surge to the chips on start up doesn’t hurt. That may be negligible but I don’t want to test it out.

Ah…..

In reply to Question,

I remember him using “constant temp” just like you used 🙂

Answer

In reply to Question,

I get home and recite the following words:

Klatu berada nichto

When I get into work the software updates are all installed. Life is easy in Trifdadoombee.

😀

Okaaaay, Mr. Wiseguy :)

In reply to Answer

🙂 what is the inflection on the words? I need to start using that. You know, whatever works is good.

Poor English

In reply to Answer

Your spelling is wrong. It is spelled Tribadoombee. What’s the matter? Don’t you have a zgrabrilator on your computer? Mine tell me when I have too many byte bugs in the RAM. TMRADIUS

But….

In reply to Question,

what about the possibility of a power surge? I know everyone should have a power surge protector but I have seen where the one piece of equipment that was off at the time the surge entered the site was the only piece of equipment that had survived.

To answer the original question whether it is good or bad to keep your pc on. I think they both have their pros and cons.

For example, if you keep your pc on, you may open yourself to security threats, thermal issues (especially if your pc is locked up in a cabinet without air flow), and the issue with memory not being released. I have found that sometimes a reboot does not clear the pc up completely you have to shut it down completely.

On the other hand, if you shut down your pc you have the expansion and contraction of all of your components that could cause a crack in your processor or any other component for that matter.

From my experience however, there is probably more risk in keeping it on than shutting it down each night. The likelihood of your components becoming damaged from turning it off is probably not that great, anyway how long are we supposed to keep a machine? If your pc lasts three years you’re doing great.

GirlGeek12,

In reply to But….

some good points. Power strips can fail or not be high enough protection.

Anyone who keeps their computer in an unvented cabinet probably deserves to have it burn up on them. Actually, I had one user in my office that had his in a closed cabinet until I happened to find it and had him put his hand on the computer. He keeps the door open now.

My office keeps the PCs until they die a very terrible death. We have had some PCs for about 8 or 9 years. It’s possible.

Thanks.

surge protection

In reply to But….

is a must. But I also recommend a UPS. Currently all of my systems are on UPS’s, as I have had problems in the past with blackouts.
a blackout (and bad shutdown) may cause HW problems of all sorts, often HDD issues. Most recent experience of a powerout was that all components on a USB hub, the hub, and the USB port on the system all died. And this was on a surge protector. I had to replace the hub, mouse, and the front panel port.

Surge Protection with Backup UPS

In reply to surge protection

doesn’t always do the trick with surges. I had a pretty hefty surge protector with battery backup in one of my closets and a cisco switch fried while it was connected along with the surge protector. If the surge comes through the ground side, nothing is gonna stop it. That’s why I’m all for shutting down pc’s before people leave. You never know what kind of weather we will have.

That’s not to say that I would get rid of my surge protectors just because one piece of hardware fried, they just aren’t 100% and people need to realize that.

Theory vs. reality

In reply to Theory vs. reality

I shut my computer down at night. I think it works better, and if there has been some small glitch or problem, that seems to work itself out also. I also don’t like leaving my wireless set up on at night, should someone come by and steal my information.

Restart versus shutdown

In reply to Does restarting desktop each day have any adverse impact?

As with most all electronic equipment, the biggest strain on the hardware is the cold start up routine. This puts the biggest strain on the power supply, system board and all parts in general. Large company/corporate businesses can achieve large cost savings by having all users shut off their machine before going home. Schedules can be set up to automatically turn their machines on prior to users arrival, and updates etc. can be done at those times. However for most home users, leaving the units on forever really does no harm, and only negligible savings would be noticed by shutting down the machine nightly. Scheduled times to due updates for both WIN and virus checkers can be done at night, and not infringe upon users times during the day. I presently have a unit that has been on continuously for two years, and nightly updates make system maintenence transparent and takes away no time from users during the active daytime hours. Daily restarts are done at the end of the day, and facilitate updates that are triggered by start up routines. It’s six of one, a half dozen of the other. Whatever is the easiest for the user, but at least a restart should be done daily.

Mj, since the

In reply to Restart versus shutdown

cold start up could have that kind of effect why would even large companies want to shut them down each night. I agree that home PC doesn’t use enough power to even realize it on your electricity bill but would it make that much of a difference in a larger corp?

Electricity cost savings of daily shutdown

In reply to Mj, since the

[i]”…would it make that much of a difference in a larger corp…”[/i]

Not as by percentage of total corporate expenses. But, a corporation with thousands of PCs will save substantial amounts of money on electricity costs. If the PC is only in use about 8 out of 24 hours, that 16 hours of running unused really adds up.

ANNUAL COST OF RUNNING A PC

According to a Microsoft article (link below):

[i]”Lab tests done by Dell show that a PC running Microsoft Office uses 42.7 watts, McCall says. If it runs continuously at that rate for 365 days, at 7 cents per kilowatt-hour, the power consumption costs would be $26.18 for the PC and $45.99 for a regular monitor, for a total of $72.17 for the workstation.”[/i]

If a large corporation has 5,000 PC workstations like this, the annual electricity costs are:

24 hours per day……….. : $360,850

minus 8 hours per day…… : $120,283

equals Savings…………. : $240,567

HERE’S A TIP

If you are working at a such company that does not already have a “daily turn off” policy, you might earn yourself a good sized bonus by recommending this and saving them a cool quarter million dollars EVERY SINGLE YEAR. I know someone who got a $10,000 bonus who did exactly that. You heard it here first. It’s spelled “T-E-C-H-E-X-E-C-2” :-).

STRAIN ON HARDWARE

IMHO, any strain on hardware caused by daily shutdown/startup is greatly overstated. It’s a myth.

REFERENCES

Do you need to turn off your PC at night?
http://www.microsoft.com/smallbusiness/resources/technology/hardware/do_you_need_to_turn_off_your_pc_at_night.mspx

edit: clarity

Your reference stated

In reply to Electricity cost savings of daily shutdown

“running Microsoft Office uses 42.7 watts.” If no programs are running and the monitor is suspended, or turned off then the amount of electricity the PC uses would be greatly reduced. Do you have any tests results on that?

Working backward from…

In reply to Your reference stated

Working backward from the figures in the article…

TURNING CRT MONITOR OFF FOR 16 HOURS PER DAY

The regular monitor (CRT) electricity cost was $45.99 out of the total $72.17 annual cost. So, leaving the PC on and turning the monitor off would be:

24 hours per day for CRT monitor…… : $45.99

minus 8 hours per day…………….. : $15.33

equals savings…………………… : $30.66

times 5000 PCs…………………… : $153,300

I doubt there would be much difference in electricity cost between running Office and not running Office. I have no figures for that.

CRT VS. LCD

As you likely know, replacing old CRT monitors with new LCD monitors will save a lot of electricity also. The MS article said regular CRT monitors use 75 watts while LCD monitors use only 22 watts. That would be:

( 22 / 75 ) * $45.99 = $13.49 operating cost per year for the LCD monitor

LCD savings is then $32.50 per year when operated 24x7x365 ($162,500 annually for 5,000 monitors operated 24x7x365).

Larger companies have a turnaround

In reply to Mj, since the

on average of 3-4 years with a new system. This is not always so, but in general. Home users often wait until their system dies completely in 3-8 years before replacing, depending on needs.

That’s policy here

In reply to Restart versus shutdown

You restart at the end of the day. Some are pushing to have users restart when they go to lunch, as well.

Restarts are essential to a healthy system. While most of the reasons are related to software, hardware issues are relevant. And, I must distinguish between a restart and a cold doot–not the same creature! A restart won’t result in a serious temperature swing in your machine–the power isn’t gone for long enough for lines to cool. If you shut the machine down at the end of the day, it really is ‘cold’ the next day when you turn it on. I’ve seldom had power supplies or fans die while a machine is running or during a restart—they tend to go on a cold boot. Says to me that the cold boot is harder on the components.

Informative post

In reply to That’s policy here

thanks.

automatic reboots nightly

In reply to That’s policy here

Our computers are now rebooted automatically at the same time every night, because many users were not logging off as instructed. Now they have to log in each day, which triggers a script which installs any new patches or system changes. A 5-minute onscreen warning is given to allow an active user to decline the reboot if they are in the middle of something.

There are some occasional problems for the folks who do not exit their applications before leaving their desks, but overall there are less problems now than before.

I would be interested

In reply to automatic reboots nightly

in finding out how you get the auto reboot. I could definitely use it here.

A few commands to know

In reply to I would be interested

I use this when I am remotely working on someones system, and they walk away before I am done.
%systemroot%\system32\rundll32.exe user32.dll, LockWorkStation
copy that entire line (including lockworkstation) and it will be like pressing ctrl-alt-delete and choosing Lock Computer

To restart a system, this command works fine
shutdown.exe -r

To shutdown a system, this should work
shutdown.exe -s

to force application closures (like end task)
shutdown.exe -f

And I was only quick enough one time with this one, to abort a system shutdown
shutdown.exe -a

All of these can be created in batch files or scripts. I use them in batch files, or copy/paste into start — run

If you’re worried about user data loss…

In reply to A few commands to know

when automatically shutting down systems, then there are various software packages available that will save data and then perform the shutdown. We are currently investigating one such product called Nightwatchman (see here for details: http://www.1e.com/softwareproducts/nightwatchman/index.aspx)

Let It Die

In reply to If you’re worried about user data loss…

In my experience, restarting solves and prevents many problems. At the company where I work they hang onto computer so long, I’m happy when one of them dies. Then I can replace it with a new one.

Careful with the auto reboot!

In reply to automatic reboots nightly

I would be careful of setting the system up with an auto reboot. Our company has a policy of running updates and performing system scans during the evening. We also have some folks that start long running applications before they go home to take advantage of the evening hours where there is less network traffic. Automatically rebooting a workstation running one of these apps would undoubtable generate a few unwanted emails.

Lunch Reboots

In reply to That’s policy here

We use Autodesk CAD 2006 products on windows xp systems and the “fix” from Autodesk for system slow down is to reboot for about every 4 hours of use. Regardless of the system or amount of RAM, Autodesk appears to slowly “eat” the memory and resources. While the users are happy we finally found a “fix” for the “lunch time crawl” as tech support we find this solution annoying and hope the next update Autodesk will resolve it. After all they have only been around for 20+ years and one would hope they can hire prgrammers that can resolve this issue. (BTW – this is a known issue since at least when 2000i came out)

Two reasons to turn off your PC

In reply to Restart versus shutdown

Dust: the longer you leave it on, the more dust is sucked in via the fans and builds up inside the case, the power supply and the processor heat sink, etc.

Surges or parts going bad: if power outage occurs when you are asleep / away, the PC will be shut down un-gracefully. (unless you have an ups and it doesn’t last longer than the UPS can carry, or you have the cable too, most non-server and home PCs do not have smart-UPSes)

I find that occasional bad disk controller, or whatever tends to trash PCs if left on for really long periods, unless you take care to clean them periodically and have UPS, surge protectors, etc.

And as other have said, you pretty much MUST reboot windoze PCs at least once a week, and for some every day. Linux probably not.

I also like to put the non-OS 2nd hard disk to power saving mode after a couple hours.

Servers versus desktops

In reply to Does restarting desktop each day have any adverse impact?

Servers that live in a controlled environment, and that contain critical information, should stay on. Servers are engineered for longer life – better power supplies, cooling, etc.Preumably any software you run on a server should be stable.

Desktops generally don’t require as long a long a life. The configurations, software wise, are fluid. You often run software that might have memory leaks.

I find with a Windows desktop OS and the kind of software I run, its better to reboot every day. Good quality desktops wont have a problem with this, hardware wise.

James

thanks.

In reply to Servers versus desktops

I reboot for the software reasons – memory leaks, scheduled updates and other pushes.

My conversation with my friend, triggered an interest on what happens to the hardware during the everyday restarts.

My interest would be little more on the theoritical side 🙂

Theoretically

In reply to thanks.

Theoretically every chip will fail eventually.

And Theoretically, the startup cycle which takes the chip from room temperature to operating temperature in a short period time stresses a chip more than running the chip at operating temp for x number of hours – I think x is an indeterminate variable, by chip.

On one hand, every motherboard has hundreds of chips/capacitors/components that typcially do not fail very often. Most outlast the usable life of the system. But if one of them fails, it takes out the whole motherboard.

I’ve seen older PCs still running after 12 years. They were never shut down.

James

Additional theory…

In reply to Theoretically

I’ve heard that the electromagnetic fields created by all the capacitors (etc.) can create stress on the various components. This stress would be generated even during a restart.

Anybody know if there?s any truth to this?

Sig: posted by former lurker trying to get involved

KHigh,

In reply to Additional theory…

I can’t answer your question but glad you are getting involved. Jump right in–the water’s fine. 🙂

No hard data

In reply to Additional theory…

I have worked in companies that, due to HIPPA requirements, required each end user to restart (not a cold boot) at the end of the business day.

In that environment, we went through a bunch of HDDs. I couldn’t initially tell what the root cause of the burn out was but discovered a couple of things.

The machines that were eating HDDs were all running Fujitsu HDDs (OEM from Compaq) and were all installed within 12 months of one another. The lions share of them had been running for approx 18- 20 months, and in that time, the restart policy had been in effect. On the other hand, I also found evidence from Compaq that they had experienced a large number of failures with that particular run of Fujitsu drives.

End of the day, I cannot say that it was the restart policy that caused the run of failures as it could have been the known flaw in the drives.

Welcome to the boards, Khigh. We’re a decent bunch…

Fujitsu Drives

In reply to No hard data

I would strongly guess that the drives where somewhere in the models of MPG3xxx somewhere around 2003/2003?

If so, that was a hardware problem with those run of drives and had nothing to do with restart policy. I heard something about red phosphorus not being completely cleaned off the controller chips during manufacturing. The chips where NOT manufactured by Fujitsu, they just were unlucky enough to use that run in their controllers.

We bought a LOT of computers (not just Compaq) during that time with Fujitsu drives in them, and have over the years had to replace ALL the drives when they died. We even got to the point that when we found a Fujitsu drive in an older computer, we immediately replaced the drive.

I understand they no longer have that problem and it was ONLY desktop hard drives. Server drives where unaffected. It pretty much killed them for awhile. No one who knew about it would buy a Fujitsu drive no matter if they knew it was from a different batch or not…

I can say that…

In reply to No hard data

We had over 150 of these pc’s with the fujitsu drives. Our endusers would do both, reboot and keep the pc’s on. Around half of the hard drives failed within a year to a year and a half. When we started to swap out the pc’s, the same variables were present, we did not have nearly as many burnouts. However, HP/Compaq put out a desktop that had a fan housing on them that distributed the air flow. We had about 15 of these in house. All of the sudden we got calls that the pc’s would not boot. Turns out the air distribution or lack there of was causing the pc to overheat. Once the pc cooled off and you pulled the housing off, the pc 9 times out of 10 never had another issue. When researching what was different about these pc’s, HP outsourced the production to Lite-On, they were not one of the regular manufacturers of pc’s.

BTW- There was a class action lawsuite again compaq for those hard drives.

Some truth, but only during startup

In reply to Additional theory…

There may be some very low level current (nano- or picoamps) induced in adjacent components during startup (cold boot OR restart). However, after startup, once the capacitors reach their normal operating condition, any EM fields around them are static.

Mr. Edit, that’s me!

Thanks for the info

In reply to Some truth, but only during startup

That’s a good enough answer for me.

btw I love you’re avatar. There are many, many days I feel just like that!

I Hibernate

In reply to Does restarting desktop each day have any adverse impact?

From Mon thru Friday I have my machine hibernate when I leave. On Friday I shut down.

This saves me the time of waiting for Windows to load. Every so often I will have to restart in the middle of the week depending on how hard I’ve been working the machine. But otherwise, it’s just fine on Hibernate. There’s an article on it somewhere (probably Microsoft.com or Help), don’t have time to find it right now.

I do note that sometimes when you close crertain programs the processes are not released completely. I usually stop them in Task Manager and I can continue working without a restart.

Tink

Oh yeah….

In reply to I Hibernate

…I shut the desktop on Fridays too 🙂

Tink: This article?

In reply to I Hibernate

Do you need to turn off your PC at night?
http://www.microsoft.com/smallbusiness/resources/technology/hardware/do_you_need_to_turn_off_your_pc_at_night.mspx

good to know

In reply to Tink: This article?

Thanks for that article. I wasn’t actually concerned with power usage. I mainly hibernate because I read somewhere that it’s unnecessary to turn off every night if you hibernate, and I HATE having to wait for my computer to load windows.

With Hibernate it only takes a few seconds for it to reload and get to the login screen. And I can leave my Outlook running so I don’t have to reset my snoozes on all the task reminders…big plus for me as I have many. 🙂

Tink

I never really added it all up

In reply to Does restarting desktop each day have any adverse impact?

But I tried to today, lets see, 166 individual locations, some with just 3 pc’s, but I came up with a total of over 6,500 pc’s. Some buildings are 6 floors with 40 rooms to a floor and most rooms have at least 2 pc’s. Each floor has a printer room with three pc’s as print controllers and 6 printers. I didn’t even think about the mainframe terminals, which are really pc’s with VM software, they are located three to a floor with the modems and controllers located in the basements. And this is just what I have to service, the corporate headquarters has twice the hardware, and three techs to do the service. That’s just here in the SF Bay area. Iknow that what they have saved in electrical bills allows them to spend more on capital expenditures, meaning more pc’s and so on. At one time, 30 years ago, they started out with 6,500 mainframe terminals, only 3,000 dumb terminals and no pc’s. Now it’s only 1500, mainframe terminals, no dumb terminals and 30,00 pc’s. That’s a big increase, but they use less power than the old equipment. The old mainframe terminals never got turned off, and most had a permanent logon banner burned onto the crt’s. Some real old stuff, and heavy too. I’m all for turning it off if it’s not going to be used within the hour. Maxtor has been the best hard drive provider with Samsung second. I cringe when I hear Hitachi or Fujitsu.

Power cycling stress on electronics

In reply to Does restarting desktop each day have any adverse impact?

Computers aren’t the only devices with electronics in them. Do you leave your TVs, stereos, and game boxes on all the time?

If you agree with the assumption that power cycling degrades component life, then you should.

The fact of the matter is that most people don’t because they pay their own electricity bill. However, they don’t see the effect of constant usage on the amount of electrical consumption where they work so they assume that it is negligible.

True, when systems suspend or hibernate, it reduces the amount of use considerably, but there is still a draw.

Just my 2 cents…

leaving household items on

In reply to Power cycling stress on electronics

New here ? yes, I turn TV?s and such off when not in use but ? I leave the microwave, refrigerator, telephone charger along with a host of other items right down to the alarm clock on 24-7. VCR?s and such I leave on too because of that flashing time and preset favorites that are a pain to program into them. Leaving on or turning off is all relevant to the application at hand.

It’s very simple…

In reply to Does restarting desktop each day have any adverse impact?

I work at a help desk for a company of 30,000 + and they require logging off only, at the end of the day. 90% of calls I receive in the mornings are resolved by simply re-booting their machine.

I have tested this theory on a certain group of people buy having them re-boot every evening or morning. that group of people have far less problems that the rest who don’t.

The slight downside to shutting down before leaving is that if Corporate pushes an update during the night, the machines will not receive it until theylogon the next morning, which causes slower response times and may require you to re-boot anyway.

rebooting in safe mode & cleanup then defrag

In reply to It’s very simple…

I have a very very small office with just three networked computers. Every night before leaving if the computer is not in use I reboot into safe mode and do a disk cleanup and then defrag the hard drive, turn the monitor off and walk away. In the morning I simply reboot the computer after closing out of defrag. It has made a huge difference the next day when running memory hog type programs! My 2 cents worth may not really apply to a huge number of computers, but the process is quite simple once one learns the sequence and who knows, it may be something that can be done automatically at a specific time in the wee hours when nobody is present.

Defrag

In reply to rebooting in safe mode & cleanup then defrag

I’m not sure why you boot to safe mode to run a defrag but you can put the defrag into scheduled tasks and schedule it to run automatically at a given time.

RE:Defrag

In reply to Defrag

That is probably a carry over from Win98. most of the time 98 would not complete a defrag without running it in safe mode.

Summary and My Thoughts

In reply to Does restarting desktop each day have any adverse impact?

Many points in this thread, some good, some less so. Part of the problem is that the observed equipment and environments vary signifigantly. I will try to classify in that manner.
1. Servers – controlled environment, high capacity/redundant cooling systems and applications that are not stopped/started frequently. Memory leakage (cause of slow down) is minimal because their applications are generally started at boot and are rarely restarted. Restarting makes a minimal impact on component temperature. Full cold shutdown (drives return to ambient temp)are hard on drives because of head stickage and congealed spindle bearing lubrication. I was a server admin for eight years.
2. Desktops – somewhat controlled environment, less cooling capacity than servers but way more than notebooks, applications started and stopped frequently. Memory leakage from starting/stopping is a major issue and the cure is a reboot as a minimum. Daily seems to be a good general recommendation but individual usage patterns may require adjustment. Thermal cycling is largely a non-issue on just a reboot but is probable no more than what is experienced in normal running (modern energy saving hardware) as the CPU varies the clockspeed, variable rate fans adjust speed in response to load. Beyond that thermal cycling failure is an early failure mode of electronic components, components that have withstood a period thermal cycling will likely survive their usefullness. On the other hand the operating life of a component is, in part related to its operating temperature, higher temprature – shorter life. Running a desktop 24 hrs. vs. 8 hrs./day will move 3 times the amount of air through the machine hence 3 times the dust buildup and higher resulting operating temperatures. I have observed this in my workplace which shut computers of at the end of the day vs. the college I taught at which ran desktops continuously. The college experienced greater HD and cooling fan failures and after three years the machines were generally “flakier” (and definately dirty.) I also have a security issue with running a desktop continuously in an enterprise environment. A machine that is shut down can not be compromised when something gets loose on the network at two a.m. and significant protection from power line fluctuations.
3. Notebooks – much lower cooling capacity than desktops, highly variable environment and frequently started/stopped applications as in desktops. Rebooting has the same usful memory leakage benefits as in desktops. Notebook computers typically run hotter than desktops thermally wearing out the electronic components faster than on a desktop. The longer the component are kept at elevated temperatures the shorter the life. Our folks that run their notebooks continuously have more problems than those that don’t. We also have about as many notebooks hard drives while in use as on boot.

My advice for desktops and notebooks. Shut them down when done or at least at the end of the day. If there is an administrative requirement to leave them on during non-duty hours at least reboot in the morning.

No

In reply to Does restarting desktop each day have any adverse impact?

and No

Daily restarting for Windows machines is usually

In reply to Does restarting desktop each day have any adverse impact?

of a benefit to the software, especially if you have set the policy settings to’ Clear VM On start Up’. No such benfit from Linux or Unix based machines.

On the hardware side, the most wear on the electronic components comes from cooling down and warming up again. So turning it off over night and back on again each morning to wear the hardware out faster. Howver, a software restart or reboot will not have this problem as it does not cool off enough to be an issue. Turn the monitor off every night and the system will use very little power over night, some systems it will use less than a start from cold will use.

Restarting a Windows machine every day or so will give you a performance benefit solely due to it overcoming deficiencies int he software by clearing the ram and VM.

Hmmm, did you mean …

In reply to Daily restarting for Windows machines is usually

… the policy setting called “Clear pagefile at shutdown“, because I’ve never heard for “Clear VM on start up” one ?!!

best regards,

Ivan Tadej, Slovenia

http://tadej-ivan.be

Yes, sorry about that – I’m thinking of what they

In reply to Hmmm, did you mean …

called it when I first saw the setting in an early version of NT. I now use SimplyMEPIS 6 and don’t have to worry about these settings any more, so I have to go from memory – can’t check it on the computer when i don’t have it. Glad you were able to find it.

Half the trouble with the system tweaks is that MS don’t tell you that they’re there or how to use them. You have to find them for yourself, or have someone tell you. If you want to go to the trouble of doing it, there are lots of tweaks and registry setting that you can use to almost double the performance (actual halve the non-performance factors) of any current MS Windows system. But it takes over an hour to do tham all, and you really need to know what you’re doing.

When I used to set up NT boxes for servers, I had a 80 page manual on how to do all this and to harden it against outside attack. Several hundred setting adjustments to give it the same basic security as an out of the box Unix/Linux system of the time. All of them were changing the default setting from ‘leave me open to attack’ to ‘protect me’ – they showed how much MS cared about security, all the basics were set wide open. And people wonder why their systems get hit.

DLLs are the biggest culprit!

In reply to Does restarting desktop each day have any adverse impact?

According to the DESIGN of Windows, once a DLL is loaded, it is not normally ever unloaded from memory. This means that other programs that use the same DLL will load faster.

It’s why, when you re-start a program you had running a few minutes ago, it starts up much faster than it did the first time. (Didn’t you ever wonder about that?)

It also means that most programs don’t unload a whole lot of memory when they are shut down, since typical modern practice is to have MOST of the program in the DLLs. Sometimes, the “.exe” file is only a few Kb, and the DLLs are tens or hundreds of Mb. They do this, by the way, for faster loading, since the program can start to run as soon as the “.exe” file is loaded.

So, what does get unloaded when a program shuts down? Only the “.exe” file and any memory the program allocated dynamically for its own storage. The vast majority of the memory used to hold the program itself stays loaded, because it’s in the DLLs.

That’s why your machine runs so much better when you restart it or shut it down. Hybernating, BTW, doesn’t help at all.

If you don’t notice a big improvement from restarting, it’s probably because the machine is always running the same applications, and you’re lucky and they don’t have big memory leaks (*ahem* web browsers *ahem*).

Either that, or you have so much memory you can afford to have gobs of it tied up in DLLs for programs that are no longer running.

To help with this

In reply to DLLs are the biggest culprit!

http://www.tweakxp.com/article36974.aspx

This will always unload a .dll when not in use. But there is still a problem with shared .dll’s

Not true about DLLs

In reply to DLLs are the biggest culprit!

The believe the DLLs remain in memory until a restart after being loaded once is complete misunderstanding of the underlying operating system dynamics.

Mapping DLLs into a process’s virtual space is similar to that for EXEs as far as the kernel’s image loader and process rundown routines are concerned.

EXE’s of course contain the map of DLLs used and specify the load address for the image loader to jump to once the overall image is mapped into the process’s virtual space – but the basic spatial mapping process involved is the same for EXEs and their DLLs.

There is no design element that permanently locks DLLs in memory once accessed. In fact, DLLs (like any code) are not even completely loaded into memory upon reference.

The virtual memory subsystem’s paging executive handles moving pages of code and data in and out of a process’s active working set as necessary.

Open DLLs ARE registered in kernel memory structures so that other processes that might subsequently access them can find and map them faster and so the paging executive can multiply map DLL pages across processes as needed. As with ALL memory pages, pages that are paged out of a process’s active working set and placed on the system’s page cache first. If that process (or another multiply mapped process) needs the page it can be mapped directly from the cache rather than paged in from disk. If memory resource demands are such that additional processes need pages that are not in the page cache and currently unassigned RAM is limited, then cache pages are recycled on a least recently used basis. Once all processes that have loaded a particular DLL terminate, the active DLL structure is deallocated (i.e., once the use count on the DLL structure hits zero, it’s deleted). However, even while some process is using a particular DLL, the DLL pages themselves are not locked in memory any more than the EXE code is.

What allows a process to load a little faster after its related DLLs have been loaded by another process is (1) the loaded DLL table provides a short cut for the image loader to map the now already open DLL file without involving the comparatively much slower disk and file subsystems and (2) some of the DLL pages are probably already actively mapped or still sitting in the page cache.

Like many other memory structures, the system does cache DLL structures for a period of time after the last process using it unloads. This is a general assumption that it might be usable again soon and thus saves involving the disk. As needed by memory demands or on a timed reclaim basis, unused DLL structures are eventually destroyed (and of course cached pages are always recycled as needed).

There are parameter settings that will force the executive to immediately dump these cached DLL structures and any related cached pages when the use count on the DLL drops to zero but unless you’re using extremely large and seldom used DLLs and your machine is already memory strained it’ll provide little or no impact – except that you’ll have the marginal emotional satisfaction of seeing an immediate (although irrelavent) change in “available” memory.

Yes, but.. . . . . (it’s MS)

In reply to Not true about DLLs

True, true, true, and true.

But evidence suggests that it doesn’t work exactly the way it’s “supposed to.”

Specifically, the release of DLL memory is, shall we say, too timid. I understand all about reference counting, but it just doesn’t seem to work as robustly in Windows for DLL memory release as it does, for example, in a Unix file system (releases file storage when # of hard links = 0).

Truth is, that’s okay, since there’s a simple solution: reboot the thing regularly.

How else can we account for the fact that in environments where I had an NT or XP workstation sitting there running the same 4 apps constantly, it would run for a week with no problem, but when I open and close a dozen different apps during the day it gets cruddier and cruddier?

For example, with my notebook (704 Mb RAM, ~500 Mb Commit Charge), if I leave it running more than about 48 hours, starts rebooting Explorer every few minutes. When I fire it up, it’s using <300 Mb; after a day, close all apps and let it sit for an hour, and it's around 400 Mb.

To me, this seems like just another example of Microsoft delivering a moderately-well implemented version of a really good idea. It seems to me that's pretty much all they do. Good implementation is apparently either too expensive, takes too long, doesn't pay, or takes people better than the ones they have doing implementation.

I really am open to other explanations, if you care to offer; I just can't seem to come up with them myself.

??

In reply to Yes, but.. . . . . (it’s MS)

Letting the machine “sit for an hour” doesn’t alter the dynamics of page caching.

Cached pages are recycled as/if needed, not on a timer. Virtual memory management still basically works like it did in the VAX/VMS days (from which Dave Cutler and crew ripped off most of the original Windows NT).

I use loads of different apps over and over again all day long on a machine with a 2GB footprint – video editing, graphics, Visual Studio, MS Office, IE, etc. and NEVER have to reboot.

On the other hand, I have seen PCs that are using crappy network or printer drivers, “acceleration” software, or just plain malware that will chew up physical memory overtime.

What OS are you running – that will make a difference

In reply to ??

When at tech college all the students had great fun in the labs Win NT 4 with MS Word, open and close Word a dozen times, with a program showing the amount of RAM and Virtual memory, and you could watch them climb, about the twelfth or thirteenth time of opening Word, the system would crash with an ‘Out of memory’ error. Can do the same thing with Win 95, Win 98, Win 2K. The problem isn’t so bad with XP as it will move it to VM, and keep expanding the VM until it runs out of disc space before crashing.

Whose setting up your machines?

In reply to What OS are you running – that will make a difference

All I can tell you is that I’m currently at a client with hundreds of XP workstations – NONE of which are being rebooted and NONE of which have any problem. In addition, my primary workstation (which is used for everything from MS Office to Visual Studio to Video Editing) was last rebooted about 3 months ago when I added a disk drive – my XP notebook is similarly NEVER rebooted (just put into standby or hibernate) – and even an old W2K workstation I use as a performance testing benchmark hasn’t been rebooted in about six mnnths.

And NONE of these machines exhibits the behavior yuo describe.

Can’t speak to the WNT since I haven’t used a machine with that in years and W95/98 aren’t even using the same architecture – so it’s moot to discuss them.

However, the condition you describe can easily result if you’re using one or more of the many crappy network/video/etc. drivers or if you’ve got some of the flaky network service software that many DSL and cable internet providers insist on pushing onto your machine.

Try seeing how the RAM and Virtual memory is growing

In reply to Whose setting up your machines?

One improvement in XP over the earlier MS Windows, is that it’ll move things from RAM to VM more readily, and will keep expanding VM while there’s space on the drive. If your system has lots of RAM and free space, you can go a long time before it finally has a hernia, but the amount of space used by VM will grew.

I wish I had DSL, remote rural phone line – can’t handle broadband at all, not even ISDN.

Edited to correct typo – seem to be dropping a lot of ‘e’s lately

VM gorwing??

In reply to Whose setting up your machines?

So are you saying that your Commit Charge Total is always sitting at or near the Commit Chart Peak – that would mean VM was going unreclaimed (not likely) or you’ve got one or more system-level components like service, driver, background process (e.g., anti-virus) etc. (that don’t ever shut down that continue to allocate more and more VM) – that happens a lot with certain defective network drivers and some bloatware from McAfee and Symantec

Remember Occam’s razor

In Windows, certain processes wouldn’t close, and some

In reply to Whose setting up your machines?

would have VM constantly growing. You can view this in the Task Manager, once you have the window set to show VM.

The problem only ever related to processes related to Windows, or other MS software. Open MS software and you see new processes open. Close the app, and some, but not all the processes it opened, will close. Run apps like MSIE and the VM keeps growing, even wehen you just leave the window open in the background. MSIE and Outlook Express are the worst culprits for this. the real beauty is OE, open it, and regardless of how you have any other service settings, it will enable and open messenger (the service), and leave it open. Close the process, and reset the service to disable, and OE will re-enable and open next time you use it. That’s why I dumped OE two years back, went to Fox Mail, uses less resources, and has higher security as well.

Don’t use McAfee or Symantec on my main machine, haven’t for years.

I’ve only ever seen the problem occur with MS apps, nothing else. Stopped using them and the situation improved. Dumped OE, dumped MSIE, dumped MS Office, and the system performed a whole lot better. But it should perform well regardless, just poor design in the OS by MS, and ditto in the apps.

Issues with not shutting down / restarting

In reply to Does restarting desktop each day have any adverse impact?

If i’m actually understanding this correctly, let me take it from the cons of not rebooting / shutting down the system. As some of the users on here have stated; applications have issues with releasing memory back to the system. And, this is true. And, is where a reboot will be necessary.

However, what needs to be discussed is the inherent problems that can occur if a user does not shut the systems down. Not speaking about servers. From what i have learned, constant disk usage can actually wear down the bearings, and other components of the hard disk. Thus, killing it in the long run (and a very long one.)

More importantly, in my opinion; in a working environment the systems should be shut down. Although people may not think of the issues which can result of leaving a computer on all night long. One of the issues which may be faced is if an attacker is to gain access to one of the end terminals, if left on. Although one may argue about larger organizations (having to turn computers on all morning can be a pain) small – medium sized organizations can possibly benefit from this.

Although attacking a computer by not shutting it down is not the issue here, i’ll try to stick to what really is. From a preventative maintenance stand point; leaving a computer on all night long only makes the computer suck up dust and other garbage cluttering the internal boards / fans. Which, in turn can diminish the power of the system (as far as speed is concerned.) And, if IT staff / Administrators don’t have good up keeping skills with their machines… This can cause certain components to fry out. As, dust is pretty much a ticking time bomb for static discharge.

Some further info, and other points, on the above

In reply to Issues with not shutting down / restarting

With a harddrive the big wear and tear is on the bearings whilst it is spinning upto and spinning down from high speed, i.e. whilst being turned off and on. Once it is at high speed there is next to no weight on the bearings, the platters sit on a cushion of air inside the drive. With a software restart, they do not wind down fully and thus there is less wear and tear. All though this benefit is reducing with some modern drives that spin down whilst the drive is not in use.

Motherboards, and other circuit boards, take a sudden jolt of electricity when first turned on, this warms the circuits. They stay warm while power is flowing through them. Again, it is the turn on and off from and to cold that does the main wear and tear. The circuits heat and cool. Again a software restart does not allow them to cool off and thus no, or very minimal effect, here.

Dust – the majority of dust in an office environment is stirred up during the day, very littlet at night.

Having the systems on during the night, allows the system admin to run updates, upgrades and virus scans at night, and minimise interference with the workers use of the systems. Minimal risk.

External attack can occur at any time of the day, and if your gateway isn’t good enough to protect the system at night, it sure ain’t gunna protect anything during the day.

These are all risk factors that affect the system, network, and hardware – they also affect usability. It is management’s job to evaluate those risks against the benefits and decide what they want to do.

As to cost savings, that depends upon what sort of power charge system you have. Some years ago, I did a full cost benefit on this for a major Asutralian military base. The way the power charges worked (for all large usage clients) was, we paid a per kilowatt charge for each kw used, we also paid an availability payment based on the maximum amount of power used at any one time during the month – the idea was that they had to make that amount available all month.

It worked out cheaper for us to have various electrical devices running extra time than to turn them on only when needed. By setting up automatic switches to stagger turning lights on between at 4.00 am and 5.00 am, instead of 5.30 am when everyone woke up, we used more power and reduced costs as it cut the sudden 6.00 am surge by 20%. the larger office environments that had computers had two clean power circuits in them. The PCs were plugged into one and the monitors (CRTS) into the other. PCs and monitors were left on all night, monitor power circuits were turned on and off by timer, off at 6.30 pm, on between 5.30 am and 6.30 am. The end result was that the total kws used went up by 7.5% but the power bill dropped by 27% because the maximum usage at any one time was cut in half.

One large corporation I worked at, had every MS Windows machine set to stay on each night, and at a set time, they did a software restart initiated by the network administrator’s PC. During the night, updates, upgrades, of apps and AV ran, as well as defrag and other maintenance tasks – then at 5.00 am every machine did a restart to flush the memory, and reset the registry, to have it ready for the day.

I do both

In reply to Some further info, and other points, on the above

depending on the situation. At home, my computer only gets rebooted when absolutely required. It never gets shut down. The downside is I forget to clean it out until something dies. So far I have been through 2 hard drives in 4 years. But neither hard drive was new to begin with. I just don?t care at home. Anything I have on that machine worth saving is backed up to a DVD-RW. At work, we shut down the computers every night. We have a lot of power surges here, and even with surge protectors, we have problems. Our servers are on 24/7 and have significantly better quality UPS and surge protection then we could afford for all our workstations. Does the shutdown affect our hardware? I would say no. We are still running an NT machine (bought new 10 years ago I think, was before my time here, but I have all the paperwork other then receipt of purchase). It takes several minutes to boot up every morning, but that?s not an issue.
I?m not really sure there is a “Correct” answer to the question of shutdown or not, merely “correct for me” answers.

You’re right in it being a horses for courses answer

In reply to I do both

I have a ten year old PC that works as my home gateway, it’s on 24/7 98% of the time (only turn it off during severe storms and when no one will be home for more than 24 hours). I also have some older machines and hardware I use for testbeds. They all still work OK, including the 386/25 and 486/50.

PCs are designed to deliver 4 to 7 years 24/7 usage, so they should last longer than the usualy 4 year commercial life cycle. A lot depends upon how they are used, and abused. I even have some 14 year old hard drives I still use for checking older PCs, the drive has various tools etc on it, the older PCs don’t recognise the multi-gigabyte hard drives.

Regarding overnight turn off costs, a lot will depend upon your power charging system. I’ve seen businesses where their only charge is kws used, then over night turn off can save money, but a busienss where you pay a peak usage rate, then over night leave on can save you money. Most PCs draw more than double their running power when first turned on, that’s to get everything warmed up and upto speed, big starting hit and then low running usage. get a few hundred doing the big hit in the same fifteen minute period and that usage spike is huge. Same with turning most lights etc one.

What about temp. stress and brittle connections/

In reply to Issues with not shutting down / restarting

Although I understand the theory behind re-booting and/or shutting down a Windows box every so often, I think the main issue has not quite been addressed.

I have been told that hard drive manufacturers measure the life of a drive in power on/off cycles, and that you may be unnecessarily shortening the life of the drive. I have also heard that the cooling down/warming up of the mainboard circuits, associated with re-starting, is not a good thing.

Given that the life of a PC may only be 3 years or so, neither of these may be of vital concern, but should at least be worth considering.

Power Cycles

In reply to What about temp. stress and brittle connections/

HD manufactures may judge life in terms of power cycles, but we need to look at long term dust/debris build up in a computer. As fans and componants get buildup on them, the thermal disipation is lessened, putting more heat into play on your componants. This can drasticly lower the life expectancy of your system, ram and cpu in particular. But once again, I blow out my case and fans about onece every 6 months, reboot as little as possible, and never turn off my computer, at home. And it runs fine, and has for several years now.

HD Life

In reply to What about temp. stress and brittle connections/

I have never encountered any drive manufacturern rate the life of a drive in power on/power off cycles. The only life expectancy rating I have ever seen published is mean time between failures (MTBF) and the testing for that generaly does not include start/stop.

Agreed

In reply to HD Life

Surely if start/stop conditions were to blame then servers that experience zero load and then massive load quickly would be extremely vulnerable

No, because the system is turned on and

In reply to Agreed

has current running through it, thus the circuits are already warmed up. The modern hard drives can handle this sort of on / off well – designed to. The older hard drives that can’t are kept running at speed, and the only thing that changes is the head will start to seek. Again all kept warm as it’s on.

not always

In reply to No, because the system is turned on and

In S3 standby power is only supplied to the memory bus and therefore the remainder of the computer can cool down.

The sort of thermal distortion that we are talking about could occur if S3 is not configured correctly resulting in the hot and cold environment we are talking about.

Yes AND No

In reply to Does restarting desktop each day have any adverse impact?

The question is an old one ranking up there with the “chicken or the egg”….

Restarting everyday has it’s benefits. Freshen up the memory, reset the drive heads, less continuous strain – especially on the hard drive, better security at night…. the list goes on.

Leaving the computer on also has it’s benefits but the most important is to wear and tear on the machine – every component, both electronic and mechanical, goes through the most stress when you start up.
As for yemperature variances and the like – this was true back in the 80’s due to differences in materials between adapter cards and the sockets. This is much less of an issue these days since materials have largely been standardized.

The answer – toss a coin. More constant wear and tear leaving the machine running or more severe wear and tear starting up every day. Over the years, it seems to balance out pretty evenly.

Save user time; restart when done

In reply to Does restarting desktop each day have any adverse impact?

One has to weigh the pros and cons for restarting against not. But in my world, it’s a pretty solution. When you leave for the day, restart the machine. No desktop OS is perfect. None of them kill off an unused process when you are done with it; _every_ time. For a user coming to sit down at a desk and work, his day goes MUCH better by not working with the same OS/Software errors that the user(s) sitting at the desk before him had. I saw a few people saying that reboots have been the solution to Help Desk calls. Being proactive and giving the next user a “fresh” machine will prevent the call in the first place (or at lease lessen the calls that lead to “reboot and call me back if it continues”).

One of those personal preferences

In reply to Does restarting desktop each day have any adverse impact?

This reply comes to you from a PC bought Nov. 1999. It has always been turned on and left on
only as long as it was necessary, (in other words it
is turned on and off at least 2 times a day to check email- sometimes more for other chores) except for a
couple of months when the hard drive started to fail.(it took a few reboots to spin up)
After a new HD, back to turn on only when needed… needless to say this is our home PC, not
business, but just to say- it hasn’t destroyed anything yet!

Fire is a great incentive!!!

In reply to Does restarting desktop each day have any adverse impact?

I used to believe that leaving it on twenty-four seven was good for the system until an over heated system completely destroyed my home and everything in it. I now shut the system down every time I leave it. I know it is a lot of trouble to reboot but I have found that I have a faster running system and fewer perifreal misreads. Actually I can’t remember the last misread. So yes, by all means, shut down when not using.

At work we are required

In reply to Does restarting desktop each day have any adverse impact?

to leave our computers on 24/7 because they run backups, updates & maintenance during the night & weekends. We are required to logoff for security reasons.

My computers at home stay turned off more than on. Between work, travel, kids, & maybe a little sleep, I really don’t have that many hours a day to use them. Although electricity consumption is a minor concern, the “always on” cable internet connection is the bigger concern. I figure I give the hackers a little less time to try to break in to my system.

Then there’s the heat issue. I’ve been using computers since C64 & Trash80’s, so I’m still stuck on the components over-heating. Power-cycling is much less stressful on electronics than heat. What if your systems cooling system stopped while you were away – ever had a motherboard to burn up? I’ve seen it happen to people I know. I also have a couple of old 486’s that were shutdown regularly, that still work.

Yeah in fact it does have an impact

In reply to Does restarting desktop each day have any adverse impact?

In one short sentence: it’s called “thermal cycling”, meaning that the power on/off cycles are damaging to a various computer’s crucial components, including, but not limited to PC’s hard-disk, CPU, graphic-card, mobo, various other inner circuits/buses, and therefore shortens a particular device’s life-time.

If you are interested, please read the “RUNNING A COMPUTER NON-STOP OR NOT” entry/article on my website’s “articles.html” page: http://tadej-ivan.50webs.com/articles.html for more information on this subject.

best regards,

Ivan Tadej, Slovenia

http://tadej-ivan.be

Redress

In reply to Yeah in fact it does have an impact

Quoting yourself hardly constitutes authority. Thermal cycling occurs at the digital semi-conductor level as the state changes from 0 to 1 and 1 to 0. This in fact is the contributer to early failure mode of semi-conductors. The metallic leads are welded to the silicon. Any welding process has a risk of hydrogen embrittlement which cause a rapid loss of strength and ductility at the point of the weld. For this reason the standard method to produce more reliable devices is to place them (after manufacture)in a circuit and operate them for 48 hours. Throw away the failures and the remaining devices are more reliable than the total lot was prior to burn-in. For an even more reliable lot of devices they can be vibrated while burning in. More initial faiures but the remaining devices will be more reliable than the as manufactured lot. This then is the thermal cycling of semi-conductor devices and the start-up/shut-down temperature changes are largely irrelavant to whats happening at the chip level.

Re: Redress

In reply to Redress

Oh well, you are or course totally right, my bad!! You see, I mis-understood/mis-read the whole thing, i.e. I was thinking about “shutting the PC down” vs. “running non-stop” and not “restarting” vs. “running non-stop” as the topic really is all about.

best regards,

Ivan Tadej, Slovenia

http://tadej-ivan.be

Commendation

In reply to Re: Redress

By the way you are to be commended for your search for knowledge in the field of Information Technology. It can be both a rewarding hobby and a stepping stone to new career opportunities. While I have been an IT professional for 15 years I started as a PC hobbyist in the early 80’s although professionally, at the time, I was a technician working with both analog and digital control circuitry.

Just be aware that there are as many “urban legends” in the IT field as any other area of life. Just because someone states something on the web or even in commercial PC publications doesn’t necessarily make it correct. Discernment and critical thinking must still be applied.

As a server administrator I really hated having to my servers go full cold because of the difficulty of getting old SCSI drives to spin back up correctly. On the other hand I routinely shut down, when not in use, the business and personal desktops/notebooks for which I am responsible and they last as long or longer than those run continuously. I am exposed to systems operated in both environment.

Re: Commendation

In reply to Commendation

Thanks for the nice and very true words, acsmith !!

best regards,

Ivan Tadej, Slovenia

http://tadej-ivan.be

Most definately no!

In reply to Does restarting desktop each day have any adverse impact?

Workstations should be turned off at the end of the day anyway if we intend to talk abut energy conservation. That obviously means it should be expected to restart the following day normally and in my experience they normally do

[Author]

Replies