[Curiosity]

Are the daily, weekly, and monthly disk maintenance tasks done in Snow Leopard automatically no matter when the computer is on, off, or sleeping?
I got the impression that one did not have to worry about the hard drive being extremely fragmented with that OS.

[AKcrab]

1. The maintenance tasks are done automatically. You don't have to to worry about them.

2. As long as you don't fill up your hard drive, you shouldn't have to worry about fragmentation.

I dont' think #1 can be debated, but regarding #2... See you on page 10.

[CharlesS]

Originally Posted by Curiosity 

Are the daily, weekly, and monthly disk maintenance tasks done in Snow Leopard automatically no matter when the computer is on, off, or sleeping?
I got the impression that one did not have to worry about the hard drive being extremely fragmented with that OS.

Who cares? Relax, and enjoy your life.

[besson3c]

Originally Posted by AKcrab 

1. The maintenance tasks are done automatically. You don't have to to worry about them.

2. As long as you don't fill up your hard drive, you shouldn't have to worry about fragmentation.

I dont' think #1 can be debated, but regarding #2... See you on page 10.

Well, the tasks are not done when the machine is asleep, but since launchd I believe the tasks are triggered based on intervals rather than absolute times, much like anacron. That is, if a daily job hasn't run in 24 hours, run it ASAP, as opposed to "run the script at exactly 6:00AM" (which requires the machine to be awake). IOW, the jobs will all run eventually providing you wake up your computer once in a while. If you put your machine to sleep during a job it should continue when it wakes up.

Fragmentation occurs in OS X, but for whatever reason this is really only a big performance drain under FAT16/32/NTFS (Windows), AFAIK.

[Big Mac]

I've never seen a serious fragmentation problem on OS X. My theory regarding fragmentation is that if you've been using a hard drive for so long it's getting seriously fragmented, you've probably been relying on that particular disk for too long, are risking failure and really should get a new hard drive soon, but I guess that wouldn't really apply to SSDs. I've also read that modern hard drives have their own internal anti-fragmentation strategies and that they may even ignore manual defrag commands from software, but I don't know how true that is.

Oh and btw, OS X supposedly defrags smaller files transparently in the background. Considering the matter further, fragmentation is almost the least important consideration regarding long term hard drive performance. What is much more important is trying to keep a good portion of a drive free because at least with standard hard drives a drive that is more than half full will start to slow down (as hard drive masters have referenced here).

[P]

Originally Posted by besson3c 

Fragmentation occurs in OS X, but for whatever reason this is really only a big performance drain under FAT16/32/NTFS (Windows), AFAIK.

Most file systems have something called a free space bitmap, a sort of "map" to which areas of the drive are free and how large each segment is. If given a file that is 500K large, they can find a segment that is slightly larger than 500K and put the file there. This means that the only way that a file gets fragmented is if it grows over time or if the drive at some point is full enough that there is nowhere that a 500K file can be put without fragmenting it.

FAT lacks that free space bitmap or any other way to quickly locate a contiguous space of a certain size. If given that 500K file, it will pick the first open block that it can find and start writing. It will not react to the fact that the block is too small until it runs out, at which point it will fragment and find a new block. This means that fragmentation will always happen over time, even if the drive is far from full, and regular defragmentation is necessary.

NTFS does not have this design flaw, and doesn't really need defragmentation any more than HFS does. NT never had a bundled defragmentation tool until Win 2k. Since Win 2k was supposed to be the OS that brought all the Win 98 users over, and they had become used to a defragmentation tool, MS licensed one and included it with Win 2k and newer OSes. It's only really useful if the drive is (or previously has been) very full.

[-Q-]

I love threads where I learn things.

[Spheric Harlot]

I agree!

P, do I assume that in something like audio recording, fragmentation *does* become an issue? I'd think that if it's completely unknown how large the files are going to be, the file system would just start writing at the first free block and fragment as it goes along.

Or is there some sort of "smart" algorithm that software can use to automatically start writing in the largest contiguous free space? (Though on second thought, that's impossible, given that it will usually have to several audio tracks at the same time.)

[besson3c]

Big Mac, I think the premise of hfs automatically defragging in the background is a myth, fwiw

[Spheric Harlot]

Originally Posted by besson3c 

Big Mac, I think the premise of hfs automatically defragging in the background is a myth, fwiw

OS X defrags small files (smaller than 20 MB IIRC) via delayed allocation, and has since 10.2.

It also defrags "slow-growing files" (such as those with data appended over time) via Hot-File-Adaptive-Clustering since 10.3.

About disk optimization with Mac OS X

This Apple Support Document was the very first hit in a Google search for "OS X automatic defragmentation".
Try it sometime - it's awesome for researching things you're not sure about but intend to commit to a publicly searchable support archive like this forum.

[CharlesS]

FWIW, the periodic scripts have nothing at all to do with disk fragmentation. They're mainly concerned with cleaning tmp files, rotating log files, updating the whatis database, etc. It's all low-priority stuff that won't hurt you even if your Mac misses a few of them (which it shouldn't, since launchd should take care of running any missed tasks once your computer wakes up).

[besson3c]

Originally Posted by Spheric Harlot 

OS X defrags small files (smaller than 20 MB IIRC) via delayed allocation, and has since 10.2.

It also defrags "slow-growing files" (such as those with data appended over time) via Hot-File-Adaptive-Clustering since 10.3.

About disk optimization with Mac OS X

This Apple Support Document was the very first hit in a Google search for "OS X automatic defragmentation".
Try it sometime - it's awesome for researching things you're not sure about but intend to commit to a publicly searchable support archive like this forum.

Delayed allocation is not running a defrag job in the background, delayed allocation is a scheme which allocates blocks in a manner such that defragging is less necessary. Maybe you could try using this awesome research tool called Google sometime?

Or better yet, stop being a dick so that we don't have to have these stupid and pointless pissing contests.

[Spheric Harlot]

Heheh.

[besson3c]

Yeah, heheh... You win the internet. Can we be done now?

[CharlesS]

Originally Posted by besson3c 

Delayed allocation is not running a defrag job in the background, delayed allocation is a scheme which allocates blocks in a manner such that defragging is less necessary.

True; however, Hot-File-Adaptive-Clustering actually is running a defrag job (for some files) in the background.

Maybe you could try using this awesome research tool called Google sometime?

Or better yet, stop being a dick so that we don't have to have these stupid and pointless pissing contests.

Was that really necessary?

[besson3c]

Originally Posted by CharlesS 

True; however, Hot-File-Adaptive-Clustering actually is running a defrag job (for some files) in the background.

Fair enough... What exactly is the difference between this and delayed allocation?

Was that really necessary?

Maybe, maybe not... I just get tired of posts like his here and elsewhere, I'd like it to stop here. My argument is that on the internet in an environment like this (and most, with exception to the Wikipedia), people are going to be right sometimes and wrong other times, and they are going to research carefully in advance at times and not so at other times, operate based on past research at times, or no research, or a misunderstanding, or third hand knowledge which was never correct to begin with, etc. Sometimes people don't know what to Google for, miss a page for whatever reason, it happens everywhere and I think it does more harm to a community to be aggressive about confronting the person about this as SH did with me as opposed to just correcting politely as you did with me here. It is no big deal to be wrong about something.

If SH is concerned about this place being an accurate archive of info for posterity sake, those finding this thread are not going to want to wade through this sort of forum drama anyway. Most people don't respond well to the pissing contests or calling somebody out, or whatever. It really isn't necessary and serves no productive purpose.

[CharlesS]

Originally Posted by besson3c 

Fair enough... What exactly is the difference between this and delayed allocation?

Hot-File-Adaptive-Clustering is automatic defrag of certain files on the disk. Delayed allocation, as you pointed out, is not.

[Spheric Harlot]

Originally Posted by besson3c 

f SH is concerned about this place being an accurate archive of info for posterity sake, those finding this thread are not going to want to wade through this sort of forum drama anyway.

This much is true.

[besson3c]

Let's move on then. I'll do a better job double checking things next time, but can we agree on aiming to be a nice inclusive environment where it doesn't matter who knows what and how often they are wrong and stuff? Maybe I was also a little overly defensive, but surely you've experienced people being brusk and rude to people on other forums over things, as if there was some sort of knowledge pecking order hierarchy, right? I hate hate hate that sort of thing...