To Jay_zhead:

SteveJL: I have tested this extensively, partitioned drives perform about 10% worse in any case I've tested right from the get go, and as the partitions get more cluttered it gets steadily worse. All the advanced defragmentaion ulitities I've tried also warn against partitioning your drive if performance is an important factor, unless you partition it to bard off the low performance area, which some people do - but I don't find that nessesary, I just make sure my drive is always at least 40% free and dynamically defrag it once in a while.

I appreciate your sharing your experiences, but do you have any links to empirical data or white-sheets, or your own testing data? I cannot find any tech-sheets advising against it, nor any reference in my Diskkeeper defragger. Thanks.

Oh, and I do understand that performance will deteriorate as any disk or partition reaches capacity (95%-ish). Thanks.

Steve the partitioning buzz kill is circumstantial.

If you take a drive and partition it (for example only)
C: OS
D: Apps
E: Data

Your performance will take a swan dive because you have 3 data active partitions on the same drive as well as pagefile usage. Your normal course of operations cause your heads to bounce back and forth accessing data. Your access time is increased and your drive life decreased.

If D:/E: were photo storage, backups, archives for any non-frequent access files you will suffer basically no hit from partitioning.

Also your performance starts degrading far sooner than 95%. Check the following graph.



Notice performance starts dipping at about 20%. The range is from 84 MB/sec to 54 MB/sec. That 30 MB/sec is a huge difference in transfer rate from one end of the disc to the other. This is also on a Raptor which is a stout performer. A more sluggish drive may be more profound and it may make the difference in usable or not (for audio/video).

With partitions this divebomb is even more noticeable. Your partition reserves a certain area. The next partition begins in area with decreased transfer rate and then the next partition even moreso. Again for example if you divided the above 74GB Raptor in 3 near equal partitions and you are losing 30 MB/sec over the disk range just from the partitioning strategy vs disk geometry you would lose about 10 MB/sec per partition before you even factor in head bounce and access time.

Back in the day, NTFS stood for needlessly tedious file system; however, I have been using it without a problem since Win2k Pro.

ORIGINAL: SteveJL

To Jay_zhead:
I appreciate your sharing your experiences, but do you have any links to empirical data or white-sheets, or your own testing data? I cannot find any tech-sheets advising against it, nor any reference in my Diskkeeper defragger. Thanks.

Sure. Here is a passage from UD manual I took a snapshot of for you:
http://www.marshdondurma.com/eng/ud_1.jpg

As for my disk benchmarks, I really can't help you there, since I did not save screenshots of those, and it's been a while since I've had a partitioned drive in my system. But they did perform worse (at least for me), honest

Jay_Zhead,

The partitioning explanation you gave was interesting in regards to fragmenting. Indeed, I was aware of the taxing duties separate partitions place on the drive heads, IF you're reading/writing to the same physical drive. It does make sense to a degree; however, does that theory still apply if you're using separate physical drives for your OS/Apps, and the other to simply write your audio data ONLY.

I would like to know if it's smarter to tell my OS/Apps drive to grab my samples & FX, or the data drive.

For what it's worth, the reason I always partitioned my drives was because of fragmentation. My theory went like this: If you have a huge play room the size of a football stadium, and you give a 4 yr. old kid all the toys in the world to play with in that stadium, and you don't clean up after him for 6 mo. to a year that's the comparison of how your hard dr will look. So, if I partition my drives much of that data may not need to be defragged for months, whereas other may need after every recording session.

That's my two cents.

KC

OK, thanks for the info guys. It would seem that partitioning itself is not a problem, what you put on the partitions is an issue. So, if one has extra physical drives, and partitions them, one should consider what goes on what partition. This makes sense.

ORIGINAL: SteveJL ~ It would seem that partitioning itself is not a problem, what you put on the partitions is an issue.

Absolutely Steve. I actually use partitions to speed up my data access.

If you think in terms of defrag it will also make sense. We spend good money on defrag tools.

Reason: to keep the heads from excessively reading/lifting/seeking/reading non-contiguous files.
Then we turn around and create non-contiguous partitions which cause the heads to read/lift/seek/read.
Totally kills any gains we might have gotten from defragging.

ORIGINAL: King Conga

Jay_Zhead,

The partitioning explanation you gave was interesting in regards to fragmenting. Indeed, I was aware of the taxing duties separate partitions place on the drive heads, IF you're reading/writing to the same physical drive. It does make sense to a degree; however, does that theory still apply if you're using separate physical drives for your OS/Apps, and the other to simply write your audio data ONLY.

See, the thing is that, as I see it, the way you described it never happens. You cannot hope that your audio data partition will be read exclusively - when you have a system partition before it. The system partition can get accessed to read data from a plugin program file, a configuration file, the registry - the list goes on - at any given time, and if you are recording 16 tracks of 24bit audio at the same time, it might just be the straw that breaks your camel's back, and you'll get a dropout. Not to mention that when you partition a system drive you peg the C drive, which will be the fastest drive, as the system drive by default, so your audio data will have to be pushed over to a SLOWER drive. That's why it is imperative to have a physical separate audio drive, that way your audio data gets the fastest clusters. Now you can if course partition that audio-dedicated drive by itself, but once again I don't see the point: for example, you want to make two drives: one for audio data and one for virtual instrument sample libraries, as drives D and E. Now you're faced with a new dilemma: one of those drives will be slower than the other. If you divide the drive equally, then the second drive will be MUCH slower than the first, by as much as 35%. So, what do you want to sacrifice? The performance of your audio tracks or of your virtual instruments? Both are very important. Both require very fast disk access... And that, in a nutshell, is why I keep my audio drive as a single partition that I regularly dynamically defrag; that makes sure that my most frequented audio data gets the best performance, and the less frequented data gets less performance. And as for backup and storage - I have a separate USB drive for that, instead of cluttering my audio drive, even if it is on a separate partition.

Whew, that was a mouthful... Good night

from a command prompt:

Convert G: /fs:ntfs

where G: is the letter of your drive to be converted.

Non-destructive.