Out of 293044 non-zero data forks total, 288828 (98.56 %) have no fragmentation.
Out of 2583 non-zero resource forks total, 2562 (99.19 %) have no fragmentation.
Not bad for a drive that hasn't been formatted since... um... well, I got the machine almost a year ago...
The worst files are all huge - iMovie clips, .cdr images, and a couple of massive .dmg files. I could just copy them elsewhere and that'd take care of it. *shrug*
Also, it's not just the defragmentation on the fly that keeps HFS+ clean. It, unlike NTFS, tries to find a block that's big enough to write the entire file into by delaying immediate writing.
Where did you get this idea? I was to understand that NTFS does employ that very same technique. When someone describes NTFS as "braindead" relative to HFS+, that gives me doubts as to where that is coming from. They both attempt to do relocated, contiguous writes whenever possible. So what makes one braindead and the other vastly superior? Given this "auto-defragged writes scenario", how does it come about where Windows machines routinely require defragmenting while OSX does not? Something there is just not adding up for me.
Thanks to the link for the fragmentation scanner- interesting and will check out...
WRT the paltry 1-2% fragmentation scores, do bear in mind the brute increases in data size and storage mediums we are dealing with. It's not uncommon to have 10-15 GB of "stuff" comprised of the core OS and closely tied apps (this isn't even including your collection of regular apps and documents). That's a lot of data storage compared to what would fill 1-2 GB of space from the OS9 days. Thankfully, most of that stays put, is read only, and never gets fragged if it wasn't to begin with. There is still the working set of stuff that gets a lot of read and write back activity in the course of a given working day. It just may seem that the working set hasn't really grown much compared to what is stored in that 10-15 GB of "stuff". So it is entirely possible to get really tiny fragmentation scores. However, if the computer is spending most of its time in that working set, and there is fragmentation within that, then the effective impact of fragmentation is possibly "larger" than that reflected in that 1-2% (hope that made sense).
80GB drive, 5GB free, for the above numbers, but your point above is taken. HOWEVER... explain the 25-30% fragmentation seen on Windows machines on a regular basis then?
Am I an expert on the detailed internals of NTFS? No.
I only know the fragmentation I see under NTFS is horrible in comparison to a similarly used HFS+ drive. Period. For *whatever* reason, the *observed* behaviour of NTFS is that of a filesystem that has been badly designed for fragmentation. Claims can be made as to the algorithms used behind the scenes (or attempted to be used), but the bottom line is simply that every Windows machine I've ever used has needed defragmentation on a regular basis, and showed real speedups afterwards for general use. Macs? Not so much. (To be read as: I've never seen double-digit fragmentation on an HFS+ drive, ever. The few times I have defragged, the benefits have been... well, non-existent.)
I would certainly agree with you that fragmentation is endemic in Windows whether it is using FAT32 or NTFS. I am just not so convinced (at this time) that it is because NTFS is inherently inferior in some way. I just wanted to make sure it was noted that NTFS did do the same write contiguous block feature that is cited for HFS+ (that is essentially one of the advantages of choosing NTFS over FAT32, other than security). So either it is something else that HFS+ does to accomplish lesser fragmentation, or perhaps it is just the nature of how Windows runs vs. how OSX runs wrt file access.
I would certainly agree with you that fragmentation is endemic in Windows whether it is using FAT32 or NTFS. I am just not so convinced (at this time) that it is because NTFS is inherently inferior in some way. I just wanted to make sure it was noted that NTFS did do the same write contiguous block feature that is cited for HFS+ (that is essentially one of the advantages of choosing NTFS over FAT32, other than security). So either it is something else that HFS+ does to accomplish lesser fragmentation, or perhaps it is just the nature of how Windows runs vs. how OSX runs wrt file access.
While I agree with Kickaha about HFS+, I take it farther. As a Mac user since System 6.0.3 and HFS, I have never seen any noticeable benefit from defragmentation software on a Mac. I have run benchmarks before defragging and afterward on my own computers. I never measured a performance benefit above noise level. When I took my current job, I became the technical support for my secretary's Mac. She had never performed any maintenance at all on the computer in the three years that she had used it. I installed Norton, repaired the hard disk including defragging it. No noticeable performance gain. I had become convinced that defragging was total waste of time ... until I defragged a Windows computer.
I would go further... defragmentation is a dinosaur relic from the '80s. Or maybe the '70s.
I was surprised when Windows NT and NTFS arrived and they still needed defragmentation. Mainstream timeshared/server operating systems had pretty much eliminated fragmentation as a problem back in the '80s. The EARLY '80s. Defragmentation seems like something from the bad old mainframe days, like manual cluster/cylinder management or preallocating maximum file sizes.
HFS+ isn't perfect by any means*, but I would be astounded if fragmentation is a real issue.
* I really wish Apple had followed through on their UFS update in Panther and switched from HFS+ to UFS as the default file system. Yes, they could have done that without losing functionality, by implementing case insensitivity and hooks for things like Spotlight in the vnode layer above the file system, like they do for network shares.
Comments
Here's my boot volume for my development machine:
Out of 293044 non-zero data forks total, 288828 (98.56 %) have no fragmentation.
Out of 2583 non-zero resource forks total, 2562 (99.19 %) have no fragmentation.
Not bad for a drive that hasn't been formatted since... um... well, I got the machine almost a year ago...
The worst files are all huge - iMovie clips, .cdr images, and a couple of massive .dmg files. I could just copy them elsewhere and that'd take care of it. *shrug*
Originally posted by bauman
Also, it's not just the defragmentation on the fly that keeps HFS+ clean. It, unlike NTFS, tries to find a block that's big enough to write the entire file into by delaying immediate writing.
Where did you get this idea? I was to understand that NTFS does employ that very same technique. When someone describes NTFS as "braindead" relative to HFS+, that gives me doubts as to where that is coming from. They both attempt to do relocated, contiguous writes whenever possible. So what makes one braindead and the other vastly superior? Given this "auto-defragged writes scenario", how does it come about where Windows machines routinely require defragmenting while OSX does not? Something there is just not adding up for me.
Thanks to the link for the fragmentation scanner- interesting and will check out...
WRT the paltry 1-2% fragmentation scores, do bear in mind the brute increases in data size and storage mediums we are dealing with. It's not uncommon to have 10-15 GB of "stuff" comprised of the core OS and closely tied apps (this isn't even including your collection of regular apps and documents). That's a lot of data storage compared to what would fill 1-2 GB of space from the OS9 days. Thankfully, most of that stays put, is read only, and never gets fragged if it wasn't to begin with. There is still the working set of stuff that gets a lot of read and write back activity in the course of a given working day. It just may seem that the working set hasn't really grown much compared to what is stored in that 10-15 GB of "stuff". So it is entirely possible to get really tiny fragmentation scores. However, if the computer is spending most of its time in that working set, and there is fragmentation within that, then the effective impact of fragmentation is possibly "larger" than that reflected in that 1-2% (hope that made sense).
Am I an expert on the detailed internals of NTFS? No.
I only know the fragmentation I see under NTFS is horrible in comparison to a similarly used HFS+ drive. Period. For *whatever* reason, the *observed* behaviour of NTFS is that of a filesystem that has been badly designed for fragmentation. Claims can be made as to the algorithms used behind the scenes (or attempted to be used), but the bottom line is simply that every Windows machine I've ever used has needed defragmentation on a regular basis, and showed real speedups afterwards for general use. Macs? Not so much. (To be read as: I've never seen double-digit fragmentation on an HFS+ drive, ever. The few times I have defragged, the benefits have been... well, non-existent.)
Originally posted by Randycat99
I would certainly agree with you that fragmentation is endemic in Windows whether it is using FAT32 or NTFS. I am just not so convinced (at this time) that it is because NTFS is inherently inferior in some way. I just wanted to make sure it was noted that NTFS did do the same write contiguous block feature that is cited for HFS+ (that is essentially one of the advantages of choosing NTFS over FAT32, other than security). So either it is something else that HFS+ does to accomplish lesser fragmentation, or perhaps it is just the nature of how Windows runs vs. how OSX runs wrt file access.
While I agree with Kickaha about HFS+, I take it farther. As a Mac user since System 6.0.3 and HFS, I have never seen any noticeable benefit from defragmentation software on a Mac. I have run benchmarks before defragging and afterward on my own computers. I never measured a performance benefit above noise level. When I took my current job, I became the technical support for my secretary's Mac. She had never performed any maintenance at all on the computer in the three years that she had used it. I installed Norton, repaired the hard disk including defragging it. No noticeable performance gain. I had become convinced that defragging was total waste of time ... until I defragged a Windows computer.
I was surprised when Windows NT and NTFS arrived and they still needed defragmentation. Mainstream timeshared/server operating systems had pretty much eliminated fragmentation as a problem back in the '80s. The EARLY '80s. Defragmentation seems like something from the bad old mainframe days, like manual cluster/cylinder management or preallocating maximum file sizes.
HFS+ isn't perfect by any means*, but I would be astounded if fragmentation is a real issue.
* I really wish Apple had followed through on their UFS update in Panther and switched from HFS+ to UFS as the default file system. Yes, they could have done that without losing functionality, by implementing case insensitivity and hooks for things like Spotlight in the vnode layer above the file system, like they do for network shares.
Neither of those are free, by any definition.
Shareware defragmentation sofware, if any exists, is a waste of the time required to download it.
In other words, you get what you pay for. Of course, if you want to load up on free stuff and have no recourse when things get hosed, have at it...