If defragging a disk was vitally important, wouldn't it follow that Apple would incorporate a defragging utility into the OS?
]</strong><hr></blockquote>
Good point. But it would make sense for Apple to include it in the OS even if it were only of circumspect importance. Windows has included backup and defrag utilities for a long time; for Macs , you've always ha to buy them third party. It;s not a bad system, seeing as Windows OSes are always many times moe expensive than Mac OSes alone, and buying the software for OS X that equals the functionality of a Windows install (including Heart and Minesweeper ) is much lower than buying the Windows package. his is especially true since OS X is Unix and a whole lot of software is available free.
Yes, but Windows systems become far more fragmented than Mac OS (at least prior to OS X). Some IT dude once told me that the original reason for defragging programs was that the fragmentation in Windows was so bad, it was considered a "bug" of sorts. So people had to buy extra software to patch their crap OS. I don't know if this is true or not, but it would go a ways towards explaining why Windows includes defraggers, while Mac OS does not.
I know a few Mac users who have never defragged their HDs, and their computers run fine. One is a beige G3 and it's still humming along in OS 9, no problems.
What you are saying about his swap file is true however, from what I've read that no fragmentation thing in unix only works if you are using UFS. If you are using HFS+ it fragments just the same because it's the same file system. Maybe someone who's more knowlegable could chime in here ( like Applenut ). I could be mistaken but I believe that's what I've read on the subject.
As far Norton Speed Disk I've always used it. When I first installed OS X Norton showed severe fragmentation and you could see it in the pattern window. When I optimized OS X got faster not slower. I generally use it once a month ( booting from the CD in X ) depending on how much stuff I've thrown away or added.
Can you use this too much? Yes most definitely. You can over use any utility that moves files around and can cause damage.
Incidently Symantic just released a downloadable free beta copy of NUM for X. It doesn't include speed disk just yet however. I've tried it and it seems to work fine so far.
I did see that mentioned a few times, but I couldn't make sence of it, so I didn't know what to quote and what not to quote. But someone did mention it, but then someone followed up and made that point seem moot. Not sure, but my point was that the concensus seemd to be (to me anyway) that in X, defragged often hurt more then it helped and really wasn't totally necessary.
Look Kid, before SpeedDisk my HD was fragged. Afterwards it was not. Over time, it became fragged again while running OS X. Therefore, fragging was happening upder OS X. It's that simple. Don't chew my ass off for making an observation.
[quote]Optimizing both de-frags (which is a HUGE speed hit) <hr></blockquote>
Do you have any evidence that this is a "huge" speed hit? I've never been able to detect ANY difference in speed after defragging, no matter how fragmented the drive (although I've never seen a Mac drive more than 15-20% fragmented).
If it's such a giant speed-hit, then evidence of this should abound. And for evidence, I mean recent evidence on modern hardware, not old 800 rpm, 100 MB HDs.
With today's hard drives, degfragging isn't even very important, since seek times are so low.
</strong><hr></blockquote>
[quote]<strong>
Do you have any evidence that this is a "huge" speed hit? I've never been able to detect ANY difference in speed after defragging, no matter how fragmented the drive (although I've never seen a Mac drive more than 15-20% fragmented).
If it's such a giant speed-hit, then evidence of this should abound. And for evidence, I mean recent evidence on modern hardware, not old 800 rpm, 100 MB HDs.
</strong><hr></blockquote>
Hmmm...
In the past, researchers have repeatedly shown that fragmentation causes a large performance hit. If anyone wants, I can dig up, scan, and post some old articles which totally substantiate this point. (There are many.) One of my favorite papers ("A Fast File System For UNIX", McKusick et al, 1984) shows how Berkeley researchers substantially reduced fragmentation in UFS (aka FFS), thereby improving disk performance by an order of magnitude (factor of ten).
Granted, many of the research articles in my possession are old. However, because of this research, both the academic and the industrial world have believed for years that fragmentation has a major negative impact on hard disk performance. So... IMHO the onus should be on the other side to provide research showing otherwise (that modern hardware deals with fragmentation well). I'm not saying that this research does not exist. I am just very interested in seeing it myself.
A quote from an older copy of a book that came with NUM : " Speed Disk improves overall performance by eliminating file fragmentation, consolidating free space and placing it at the end of the disk, and marking off bad blocks.
Speed Disk also optimizes the placement of files on a disk so that infrequently modified files, are moved to the beginning of the disk. Since you rarely modify system files, it makes sense to place them together in one contiguous chunk. Frequently modified files ( such as a Desktop file, which is constantly updated ) are best placed at the end of the disk, to give them room to grow and shrink without causing fragmentation to the other files. "
When you modify a file it changes in size. since the file already has space allocated to it the drive has no choice but to put the " new " part of the file where it can find free space. This is called fragmentation. Kind of like when a city plans out it's streets overtime part of a street can be here and the rest of it can be several blocks away ( much like some parts of Portland Oregon ). And when this causes you to go blocks out of your way to find the rest of it, the time to travel the entire length of the street is longer. It's the same with files, the head has to take more time to find the rest of the file.
With optimization it places like files together for faster access. Files that are used together with other files aren't far apart and the head doesn't have to go searching all over the place for them.
Both discontiguous and fragmented files can slow your drive down over time.
NUM recomends ( like what I've been doing ) checking your drive with Speed Disk once a week for the user who is frequently editing or saving large files. Once a month for the casual user.
Sorry it took me so long to delve into this but, it's Sunday ( since I have to work on Saturdays ) and I have the time now.
I wish I could find the acticle about UFS vs HFS+ but I distinctly remember reading that HFS+ will fragment just like before ( because of the physical nature of the file system ) no matter what OS is on it.
Also I can vouch for the fact that the Windows system fragments more frequently because this is what happens to my computer at work.
Oh by the way, no matter how fast your modern HD is the seek time will always be less if it's less fragmented or more optimized. Given the size that some files are getting to be nowadays this is a serious consideration.
<strong>Why do you need to defrag X? I believe that that is unneccessary. Unxi files never move, there isn't reshuffling so I don't think you need to worry about defragging X ever again.</strong><hr></blockquote>
OS X fragments my hard drives more than OS 9. I don't see how you can say it doesn't fragment at all.
Is XP any better at fragmentation than previous versions? We still use 95 in our dept. at work (yuk ). <img src="graemlins/oyvey.gif" border="0" alt="[No]" />
Didn't you read the quote from the NUM manual? It all depends on what ( or how much ) you are adding or deleting from the disk.
If Applenut is adding, deleting, or moving lots files around for what ever reason he would have to defrag or optimize ( and yes I know they are two different things as I have previously explained ) more often. It depends on the individual. Ether one could have an affect on the speed of his seek time.
If your disk looks like a " complete disaster area " I wonder what program or pattern you are using to optimize. I always use the " general use " pattern from Speed Disk in Norton System Works. It never looks messy.
As I said before Norton recomends once a week for the heavy user, once a month for the casual user.
First of all, there is no magic in OSX. just because it's a new os, it's still HFS+, and it becomes fragmented just like every other fs (except for something like ffs, which is nice when coupled with softupdates)
jimmac:
the problem with windows 95 and that file fragmentation was that the pagefile was very "busy", and that caused many file fragmentations. and i would say that page file fragmentations hurt performance the most...
also, during the defragmentation process of windows 9x, it has to restart every time the is a change on the disk. (that's why sometimes defrag takes forever it seems)
With windows 2000/xp, the file defragments in stages and it has built in hooks in the api to safely defragment. also, it doesn't have to restart every time while it's defragmenting and the swapfile's a lot more stable (meaning it doesn't change in size, i personally set the swapfile to a specific size). But it still gets fragmented, and i've seen it get terribly, terribly fragmented (but only once). It took multiple defrags to clear everything up.
anyhow, if you are always reinstalling the os, or if you don't install/uninstall apps a lot, then you'll probably never see a problem. but if you have your os installed for a while and are always upgrading software, well then perhaps you'll see the difference.
Ted
also, airsluf, i think you did the calculations wrong. with rpms of hard drives, first you divide by 60 seconds (so 120 rounds per second) then you divide that by 1000 (so it'll spin .12 rounds per ms, which is a pretty far cry from the 7.2 rounds that you had stated)
<strong>What did you suggest - TechTool? I'll think about that too. Does it defragment drives?</strong><hr></blockquote>
Yep. It does basically everything NUM and NAV do, does them better and has a few extras as well. Historically TechTool is a much more reliable and efficient way to tune up your Mac.
In the space of an hour (depending on your drive size and what's on it) you can run perhaps two dozen important hardware diagnostics, checking everything from RAM and HDs, to video cards and network ports. You can also check for software conflicts (like Conflict Catcher albeit it's done behind the scenes not with a GUI) and check for viruses as well.
It's an excellent, excellent application and the support is generally first rate. Puts anything Norton has released to date, to shame and worse.
Comments
<strong>
If defragging a disk was vitally important, wouldn't it follow that Apple would incorporate a defragging utility into the OS?
]</strong><hr></blockquote>
Good point. But it would make sense for Apple to include it in the OS even if it were only of circumspect importance. Windows has included backup and defrag utilities for a long time; for Macs , you've always ha to buy them third party. It;s not a bad system, seeing as Windows OSes are always many times moe expensive than Mac OSes alone, and buying the software for OS X that equals the functionality of a Windows install (including Heart and Minesweeper
I know a few Mac users who have never defragged their HDs, and their computers run fine. One is a beige G3 and it's still humming along in OS 9, no problems.
<strong>Kid Red,
What you are saying about his swap file is true however, from what I've read that no fragmentation thing in unix only works if you are using UFS. If you are using HFS+ it fragments just the same because it's the same file system. Maybe someone who's more knowlegable could chime in here ( like Applenut ). I could be mistaken but I believe that's what I've read on the subject.
As far Norton Speed Disk I've always used it. When I first installed OS X Norton showed severe fragmentation and you could see it in the pattern window. When I optimized OS X got faster not slower. I generally use it once a month ( booting from the CD in X ) depending on how much stuff I've thrown away or added.
Can you use this too much? Yes most definitely. You can over use any utility that moves files around and can cause damage.
Incidently Symantic just released a downloadable free beta copy of NUM for X. It doesn't include speed disk just yet however. I've tried it and it seems to work fine so far.
[ 01-26-2002: Message edited by: jimmac ]</strong><hr></blockquote>
I did see that mentioned a few times, but I couldn't make sence of it, so I didn't know what to quote and what not to quote. But someone did mention it, but then someone followed up and made that point seem moot. Not sure, but my point was that the concensus seemd to be (to me anyway) that in X, defragged often hurt more then it helped and really wasn't totally necessary.
Do you have any evidence that this is a "huge" speed hit? I've never been able to detect ANY difference in speed after defragging, no matter how fragmented the drive (although I've never seen a Mac drive more than 15-20% fragmented).
If it's such a giant speed-hit, then evidence of this should abound. And for evidence, I mean recent evidence on modern hardware, not old 800 rpm, 100 MB HDs.
<strong>
With today's hard drives, degfragging isn't even very important, since seek times are so low.
</strong><hr></blockquote>
[quote]<strong>
Do you have any evidence that this is a "huge" speed hit? I've never been able to detect ANY difference in speed after defragging, no matter how fragmented the drive (although I've never seen a Mac drive more than 15-20% fragmented).
If it's such a giant speed-hit, then evidence of this should abound. And for evidence, I mean recent evidence on modern hardware, not old 800 rpm, 100 MB HDs.
</strong><hr></blockquote>
Hmmm...
In the past, researchers have repeatedly shown that fragmentation causes a large performance hit. If anyone wants, I can dig up, scan, and post some old articles which totally substantiate this point. (There are many.) One of my favorite papers ("A Fast File System For UNIX", McKusick et al, 1984) shows how Berkeley researchers substantially reduced fragmentation in UFS (aka FFS), thereby improving disk performance by an order of magnitude (factor of ten).
Granted, many of the research articles in my possession are old. However, because of this research, both the academic and the industrial world have believed for years that fragmentation has a major negative impact on hard disk performance. So... IMHO the onus should be on the other side to provide research showing otherwise (that modern hardware deals with fragmentation well). I'm not saying that this research does not exist. I am just very interested in seeing it myself.
Speed Disk also optimizes the placement of files on a disk so that infrequently modified files, are moved to the beginning of the disk. Since you rarely modify system files, it makes sense to place them together in one contiguous chunk. Frequently modified files ( such as a Desktop file, which is constantly updated ) are best placed at the end of the disk, to give them room to grow and shrink without causing fragmentation to the other files. "
When you modify a file it changes in size. since the file already has space allocated to it the drive has no choice but to put the " new " part of the file where it can find free space. This is called fragmentation. Kind of like when a city plans out it's streets overtime part of a street can be here and the rest of it can be several blocks away ( much like some parts of Portland Oregon ). And when this causes you to go blocks out of your way to find the rest of it, the time to travel the entire length of the street is longer. It's the same with files, the head has to take more time to find the rest of the file.
With optimization it places like files together for faster access. Files that are used together with other files aren't far apart and the head doesn't have to go searching all over the place for them.
Both discontiguous and fragmented files can slow your drive down over time.
NUM recomends ( like what I've been doing ) checking your drive with Speed Disk once a week for the user who is frequently editing or saving large files. Once a month for the casual user.
Sorry it took me so long to delve into this but, it's Sunday ( since I have to work on Saturdays ) and I have the time now.
I wish I could find the acticle about UFS vs HFS+ but I distinctly remember reading that HFS+ will fragment just like before ( because of the physical nature of the file system ) no matter what OS is on it.
Also I can vouch for the fact that the Windows system fragments more frequently because this is what happens to my computer at work.
[ 01-27-2002: Message edited by: jimmac ]</p>
[ 02-05-2002: Message edited by: jimmac ]</p>
<strong>Why do you need to defrag X? I believe that that is unneccessary. Unxi files never move, there isn't reshuffling so I don't think you need to worry about defragging X ever again.</strong><hr></blockquote>
OS X fragments my hard drives more than OS 9. I don't see how you can say it doesn't fragment at all.
I also see a speed boost when I degrag the drives
Is XP any better at fragmentation than previous versions? We still use 95 in our dept. at work (yuk ). <img src="graemlins/oyvey.gif" border="0" alt="[No]" />
If Applenut is adding, deleting, or moving lots files around for what ever reason he would have to defrag or optimize ( and yes I know they are two different things as I have previously explained ) more often. It depends on the individual. Ether one could have an affect on the speed of his seek time.
[ 01-27-2002: Message edited by: jimmac ]</p>
As I said before Norton recomends once a week for the heavy user, once a month for the casual user.
[ 01-27-2002: Message edited by: jimmac ]</p>
jimmac:
the problem with windows 95 and that file fragmentation was that the pagefile was very "busy", and that caused many file fragmentations. and i would say that page file fragmentations hurt performance the most...
also, during the defragmentation process of windows 9x, it has to restart every time the is a change on the disk. (that's why sometimes defrag takes forever it seems)
With windows 2000/xp, the file defragments in stages and it has built in hooks in the api to safely defragment. also, it doesn't have to restart every time while it's defragmenting and the swapfile's a lot more stable (meaning it doesn't change in size, i personally set the swapfile to a specific size). But it still gets fragmented, and i've seen it get terribly, terribly fragmented (but only once). It took multiple defrags to clear everything up.
anyhow, if you are always reinstalling the os, or if you don't install/uninstall apps a lot, then you'll probably never see a problem. but if you have your os installed for a while and are always upgrading software, well then perhaps you'll see the difference.
Ted
also, airsluf, i think you did the calculations wrong. with rpms of hard drives, first you divide by 60 seconds (so 120 rounds per second) then you divide that by 1000 (so it'll spin .12 rounds per ms, which is a pretty far cry from the 7.2 rounds that you had stated)
<strong>What did you suggest - TechTool? I'll think about that too. Does it defragment drives?</strong><hr></blockquote>
Yep. It does basically everything NUM and NAV do, does them better and has a few extras as well. Historically TechTool is a much more reliable and efficient way to tune up your Mac.
In the space of an hour (depending on your drive size and what's on it) you can run perhaps two dozen important hardware diagnostics, checking everything from RAM and HDs, to video cards and network ports. You can also check for software conflicts (like Conflict Catcher albeit it's done behind the scenes not with a GUI) and check for viruses as well.
It's an excellent, excellent application and the support is generally first rate. Puts anything Norton has released to date, to shame and worse.