Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Ever since my time of windows 3.1 i always see myself waiting the hard drive activity to stop before opening a program or beginning a batch activity. In those times, i'm sure this helped, you could clearly see that opening 2 programs at the same time took much longer than opening one, and after that was complete, opening the other. I also tend to wait a program fully close (swap out) before opening another.

But today cache algorithms and hard drive performance and caching changed a lot.

So the question is, do i still am doing good by waiting the hard drive activity to stop, or could i just do whatever i want and don't worry about it?

I ask this because my latest lenovo notebook don't have the HDD led, so i don't have a clue when it's working or not.

btw, my current system is windows 7.

share|improve this question
    
You're not hurting the drive or anything. Unless you experience a slowdown, you can just do whatever you want with your drive. –  slhck Aug 17 '11 at 9:26

4 Answers 4

up vote 2 down vote accepted

The short answer is, it depends on why the hard drive light is on. The long answer:

But today cache algorithms and hard drive performance and caching changed a lot.

Yes, this is true, but that's not why you waited for the HDD LED to stop (or rather, this isn't the reason that you don't notice the performance degredation as much anymore). Hard drives are big, mechanical devices which are good at one thing - only doing one thing at a time. You can only read/write to a single sector at a time, so attempts to use the hard drive simultaneously usually result in thrashing.

So the question is, do i still am doing good by waiting the hard drive activity to stop, or could i just do whatever i want and don't worry about it?

Again, it depends on what the HDD is doing. Why? Well, if you're transferring gigabytes of files to/from your hard drive, then you will most certainly experience disk thrashing (and this will also cause the drive head to move back and forth rapidly, decreasing the drive's lifetime). However, if you don't know why the HDD LED is flashing, then it is probably a background process.

Most background processes (search indexers, disk defragmenters, the .NET runtime optimizer, etc...) do not cause thrashing for two reasons. The first is that they run at a lower priority, so the operating system places a lower priority on their I/O requests. The second is that most of these background processes either stop or reduce their activity when your CPU or disk usage reaches a certain threshold, to avoid impacting your perceived performance of the system (or cause thrashing).

I ask this because my latest lenovo notebook don't have the HDD led, so i don't have a clue when it's working or not.

Don't worry about it then. Most background services in Windows 7 are aware of the issues I outlined above, so you will not experience any slowdowns. Unless you are trying to use the hard disk simultaneously (i.e. transferring a lot of files while trying to load multiple programs at once), you should not experience any thrashing.

Lastly, if you want to know what transfer rate a particular process is using your hard drive, see the question How can I view what percentage of my hard drive bandwidth is currently being used?


One last thing to note, having a solid-state drive totally changes the disk thrashing game. While it cannot be totally avoided (yes, even with a SSD), it will be a lot less noticeable with a solid state drive versus a mechanical hard drive (see What are the pros and cons of a solid-state drive? if you want to learn more).

share|improve this answer
    
Practically speaking, seeking does not decrease a drive's lifetime, it is meant to be able to do a lot of random seeking. Even under worst case load, the drive is still going to last years. Also background tasks might cause less thrashing due to IO priority scheduling algorithms, but they certainly do still cause some; you are, after all, still doing two things at once. –  psusi Aug 17 '11 at 17:41
    
@psusi drives are rated for an (average) maximum write cycle lifetime. Moving the drive head back and fourth in rapid succession (like what happens when an OS tries to perform two HDD-intensive tasks simultaneously) will severely degrade a drive's lifespan (for a mechanical or solid state drive). –  Breakthrough Aug 17 '11 at 18:14
    
@breakthrough: yet there is no conclusive evidence of that. Heck, back before there was NCQ, drives still lasted a good five to 6 years. And they were under more stress then compared to now. Going without a page file was inconceivable then. . . –  surfasb Aug 18 '11 at 5:16
    
@surfsab So you're telling me that if I have a disk that is constantly thrashing, and a disk that's just spinning idle, they will both have the same MTBF? I highly doubt that. –  Breakthrough Aug 18 '11 at 10:28
    
@Breaktrough SSDs are rated to have a limited number of write cycles, HDDs are not. Theoretically the more seeking the more mechanical wear on the drive, but the degree is so small that manufacturers do not specify it and in practice, nobody has ever been able to measure it. The slight difference in this discussion is not going to have any effect. –  psusi Aug 21 '11 at 19:09

It depends. If you are running two tasks at once that are doing nothing but disk IO, then they will slow each other down, so you are better off doing them one at a time. Most tasks involve at least some cpu time, during which the disk is free to work for the other task, so you can see better overall performance running two tasks like that, since while one is using the cpu, the other can do IO, and vice versa.

With regards to write activity, most of the time, writes sit in the cache for a bit, and are gently flushed to disk in the background. Waiting for all of those background writes to finish before running a new program is a waste of time, since it doesn't hurt anything for those writes to hang out in the cache a while longer while the new program is loaded first, and then when that program either goes idle waiting for you to do something, or starts doing computations instead of disk IO, then writes can finish, using otherwise idle disk time.

On average, most of the time you see a net gain doing two things at once, so as a general rule, no, you should not bother waiting for disk IO to stop before doing something else.

share|improve this answer
    
+1 for considering modern technology. –  surfasb Aug 18 '11 at 5:17

It is the job of the harddrive controller and the OS to manage Read/Write requests, not yourself.

In the old days, ya, applications would freeze up under lots of I/O. But that was back in the glory days of Software development where the C language and a text editor could write pretty much anything.

Now is the Era of Hardware. Hardware has far outpaced software development and thus today's hardware can pretty much run 100% and not miss a beat. On top of that, developers have learned their lesson. More and more programs spin their UI on another thread, thus mitigating the unresponsiveness of yester-year's Hourglass/SpinningBeachBall of Death.

share|improve this answer
    
It's actually spinning long running processes on another thread :) Certainly with Windows the UI thread is the main thread. –  ChrisF Aug 17 '11 at 10:49
    
the big problem is, hdd is not yet multi threaded –  bortao Aug 17 '11 at 23:29
    
But it doesn't damage nor crash at peak loads, does it? –  surfasb Aug 18 '11 at 5:13

On any other system it wouldn't matter. But Windows can get itself into a thrashing situation (or a deadlock) fairly easily if too much stuff is going on at once. This is especially true right after booting.

You don't have to wait for all activity to stop, but you should generally hold off until the system is responding fairly normally.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.