How often do you guys defrag your drives

[steingra]

Hello

I have always wondered but never actually tried to run any performance tests on a system that places the Windows swap file on it own disk. Whether IDE/SATA/or SCSI. With newer motherboards you can stick lots of SATA drives in there and also IDE drives. But I dont normally add an IDE drive into a system with SATA drives in it.

I would like to believe you get at least a 5 or 10% performance boost in overall Windows OS performance by doing that (adding windows swap file to its own dedicated channel and dedicated hard drive). Just a gut feeling though...no facts to back it up at this point.

Anyone do this type of a setup and notice a worthwhile difference?

Im thinking of restructuring a computer system something like this:
C Drive -> SATA II Drive -> SATA Primary Channel -> 7200 RPM
Holds the OS only.

D Drive -> SATA II Drive -> SATA Secondary Channel -> 7200 RPM
Holds the Windows swap file

E/F/G/H Drives -> SATA II Data drives in 64k block sizes for holding videos -> 7200 RPM or SATA I - 10000 RPM Raptors.

[opy01]

steingra:

I actually do this on all PCs with multiple HDs. I do notice a difference too though I havent run performance tests to see exactly how much of an increase I get. I also never put IDE HDs in with SATAs if I can help it, of coarse I don't use IDE HDs anymore. I have done that once though and copied files over a gigabit network to the IDE and SATA. The IDE took 30 minutes to write to and the SATA took 5 minutes to write to.


stanger89:

I guess my response earlier was a bit... fragmented (no pun intended). I am worried about drive longevity also. Thats why I set up a RAID, to take stress off of one drive and disperse it between several, not to mention the redundancy also. I also put the paging file onto a seperate drive to take stress off the OS drive.

I was not trying to associate noise with the amount of power consumpsion or heat caused by the drive. I was saying that my drives in that PC are very quiet so I don't hear them, then I went on to a different point about how they don't take that much power. There are several ways to cut down on heat and noise and the added storage and performance increase is worth the little extra power consupsion.

I guess what goes on in my head and what gets typed are two different things. I have the same problem talking, people give me funny looks sometimes even though what I want to say makes perfect sence (to me). But since everyone has different opinions and preferences I guess we could go back and forth indefinately.

[rickgillyon]

Originally Posted by opy01

The 200GB disk is the smallest one I have. I always found creating different partitions pointless since the read heads have to read the data from the same platters anyways.

No, because having a separate partition allows you to format it appropriately and allows you to avoid fragmentation on your system drive, avoid using a system drive for compression jobs. Why impact your system drive unnecessarily?

Also, if it's a PVR, just stick enough RAM in it (2GB) and disable the pagefile completely, will perform better.

[stevech]

I guess my response earlier was a bit... fragmented (no pun intended). I am worried about drive longevity also. Thats why I set up a RAID, to take stress off of one drive and disperse it between several, not to mention the redundancy also. I also put the paging file onto a seperate drive to take stress off the OS drive.

I subscribe to the idea that the life of a disk drive is, absent vibration and excessive temperatures, determined by motor-on time and less so by the number of seeks. Over the many, many years I've been into computers (my first hard drive was 5MB, 5-1/4 in. form), I've had quite a few fail. Virtually all were motor/bearing failures. A few head crashes, but those are increasingly rare. So in RAID5, as the number of drives increases, the full-array MTBF degrades fast. There's a point of diminishing returns, given the many reports that the array did not heal itself as promised.

[opy01]

Originally Posted by stevech

I subscribe to the idea that the life of a disk drive is, absent vibration and excessive temperatures, determined by motor-on time and less so by the number of seeks. Over the many, many years I've been into computers (my first hard drive was 5MB, 5-1/4 in. form), I've had quite a few fail. Virtually all were motor/bearing failures. A few head crashes, but those are increasingly rare. So in RAID5, as the number of drives increases, the full-array MTBF degrades fast. There's a point of diminishing returns, given the many reports that the array did not heal itself as promised.

Thats odd since the majority of the drive failures I have had were the heads. I think only 2 of them just stopped spinning, the rest would spin up but make funny clicking noises when trying to read data.

I am not trying to make a statement but rather wonder out loud, I thought the spinning up of a drive was harder on the drive motor than the motor staying running alot longer. But then you could ask the question "does leaving your lights on all day instead of turning them on and off as needed save energy?"

[stevech]

YMMV they say. Point being: disk drives are mechanical beasts: head that move and hopefully fly at a few microns; motors with bearings spin disks that are heavy as compared to the motor's torque capacity.

More disks = poorer MTBF. One can do the statistics.

RAID5 promises to deal with this by tolerating a drive failure - though many folks with inexpensive RAID5 report, by my reading, that it's far less than certain that the recovery will succeed.

RAID1 simply mirrors, so the statistics, I say, are in your favor, for a small (like < 1TB) stores.

But, that's just my opinion.

[opy01]

Originally Posted by stevech

YMMV they say. Point being: disk drives are mechanical beasts: head that move and hopefully fly at a few microns; motors with bearings spin disks that are heavy as compared to the motor's torque capacity.

More disks = poorer MTBF. One can do the statistics.

RAID5 promises to deal with this by tolerating a drive failure - though many folks with inexpensive RAID5 report, by my reading, that it's far less than certain that the recovery will succeed.

RAID1 simply mirrors, so the statistics, I say, are in your favor, for a small (like < 1TB) stores.

But, that's just my opinion.

and a good opinion it is. I have had good luck so far recovering from drive failures on a RAID 5 systems. I have also however heared of and entire RAID failing causing complete data loss (the idiots stored their backup on the RAID lmao). Of coarse no matter what your setup, your still relying on technology which is going to fail for whatever reason and sometimes theres no telling why.

[jgs14]

Originally Posted by rickgillyon

Also, if it's a PVR, just stick enough RAM in it (2GB) and disable the pagefile completely, will perform better.

Thats a myth. Modern versions of Windows like to allocate a huge address space per process. Most of this allocated address space never gets used, but it has to be mapped to some part of memory -- mapping to virtual memory allows windows to keep your RAM free for actually doing stuff (rather than just sitting there as an un-used allocation).

(The pagefile thing was true in Windows 95/98, but never for the NT based systems)

...to keep on topic, I use Windows command line defrag in a batch file from task manager to defrag my local drives every night. Every night may seem compulsive, but if you do it every day it only takes 10-20 minutes a day (and at a time when I am sleeping anyways).

--
Joe

[Jetracers]

Everyday, just run diskeeper 2007 or up and it automatically defrags your drive when there aren't heavy periods of use. It will make your system much more peppy aftewards if you have never done a defrag