at what % do I defrag?

[sbrads]

Here's some facts for the doubters....
Realities of Hard Drive Fragmentation | Overclockers

 

[JimLewandowski]

Here's some facts for the doubters....
Realities of Hard Drive Fragmentation | Overclockers

Note the test was done with a 250GB hard drive. VERY tiny by today's standards and it WILL skew the results to show defragging is better since the "walk" across the disk to inner cylinders will occur more quickly vs. a 750GB (as an example) drive.

When HDtune is testing sequential read speed, is it using REAL data (vs. just swiping a bunch of contiguous tracks/cylinders? When it shows average access time, avg access of WHAT? Random? REAL files that are or aren't in fragments?

Something is clearly amiss in the test results.

A fresh install had 39 Mb/sec read speed and 25 ms avg access time. We've got some "holes" as any updated module will have been deleted from the original fresh-install contiguous and the updated modules now lie

The 2nd benchmark (after purposeful fragmentation), the average read speed did drop but the average access time dropped in HALF from a fresh install (+ updates). How can this be possible?

Then the 3rd benchmark (aftrer a XP defrag) had the read speed upped to 50 Mb/sec and avg access time down to 7.5 ms.

Hence, the test is suspect and we don't even know WHAT HDtune does when it does these tests.

http://www.maximumpc.com/article/features/25_most_popular_windows_tips?page=0%2C2

"
Defragmenting Your Hard Drive Improves Performance: TRUE


One of the most venerable suggestions for improving disk performance is to defragment your hard drive regularly. The science of defragging is sound: By putting all the bits of a file or application in sequential order on your drive, the drive should have to do less work (and spend less time) to access those files. Thus: faster performance. Well, in practice it’s not really true. Today’s hard drives are fast enough to make fragmentation largely irrelevant, and our benchmark tests have repeatedly borne this out: On moderately fragmented drives, defragmentation will offer negligible to no performance increase. For seriously fragmented drives (think 40 percent or more), especially those running XP or older OSes, defragmentation can help, but don’t expect the world. As for third-party defrag tools, there’s no real evidence that they’re any more effective than Windows’ built-in defragger.
Click Disk Defragmenter under Accessories / System Tools.
"

 

[sbrads]

Here's some facts for the doubters....
Realities of Hard Drive Fragmentation | Overclockers

Something is clearly amiss in the test results.

I agree it's flawed, the problem is there doesn't seem to be any other worthwhile tests done on whether defragging works or not. I use MyDefrag regularly now with Win7 because I've proved to myself it helps keep a system snappy even when it's got hundreds of programs loaded over a few years (in XP anyway with JKDefrag, MyDefrag's forerunner).

The final proof (to me) was last year when my daughter told me her 2yr old XP laptop was taking ages to load Firefox and had been getting steadily worse for a few months. In actual fact it was taking ages on everything but Firefox was never the quickest program to load up so she noticed that more so. I defragged using the Windows app and it sped up by perhaps 10%, if that. I then defragged using JKDefrag, and everything loaded up 3 times quicker and that's not an exaggeration.

 

[JimLewandowski]

Here's some facts for the doubters....
Realities of Hard Drive Fragmentation | Overclockers

Something is clearly amiss in the test results.

I agree it's flawed, the problem is there doesn't seem to be any other worthwhile tests done on whether defragging works or not. I use MyDefrag regularly now with Win7 because I've proved to myself it helps keep a system snappy even when it's got hundreds of programs loaded over a few years (in XP anyway with JKDefrag, MyDefrag's forerunner).

The final proof (to me) was last year when my daughter told me her 2yr old XP laptop was taking ages to load Firefox and had been getting steadily worse for a few months. In actual fact it was taking ages on everything but Firefox was never the quickest program to load up so she noticed that more so. I defragged using the Windows app and it sped up by perhaps 10%, if that. I then defragged using JKDefrag, and everything loaded up 3 times quicker and that's not an exaggeration.

The more antiquated your system (hardware and OS/XP), the more defrag WILL show some improvement. For example, a 5400 RPM drive (laptop?) will show larger improvements because 5400 RPM is so slooooooooooooow. So, your daughter's situations IS valid.

The sharper websites have done some tests (but I can't find the articles) and concur with what I know to occur.

People still seem to be missing what Windows (even XP) does at startup to make somewhat moot, defragging. It keeps track of all I/Os and schedules them (on each successive boot) such that it minimizes the delay from disc seeks (head movement).

 

[smarteyeball]

I personally use Auslogics defrag and 'optimize' option every now and then (not the best app, but far from the worst) more for the placebo effect of having files organized contiguously, rather than for any noticeable performance difference.

One question though Jim, what if any, 'noticeable' impact does defragging have on a larger drive 7200rpm drive ie 640, 1TB partitioned into smaller partitions have? ie 100/200GB

 

[JimLewandowski]

I personally use Auslogics defrag and 'optimize' option every now and then (not the best app, but far from the worst) more for the placebo effect of having files organized contiguously, rather than for any noticeable performance difference.

One question though Jim, what if any, 'noticeable' impact does defragging have on a larger drive 7200rpm drive ie 640, 1TB partitioned into smaller partitions have? ie 100/200GB

Prerequisite: I can't find the specs for small vs. large drives to see how many PLATTERS (disc surfaces) are involved. There are performance implications on how many heads vs. # disc surfaces, etc for a given size GB footprint. IOW, rather than have only 1 head servicing a platter the size of a 12" pizza, having 4 heads servicing 4 platters each being 3.5" is vastly faster as you reduce all seek times by 1/4 with the smaller surface w/ a greater # of heads. So, I don't know if a 500GB drive has 2 heads (250GB per head) and then when you go to 750GB you now have 4 heads (180GB per head). The latter will perform better. As I said, I can't find these specs on Seagate or Western's website.

FWIW, when I bought my hard drive, since I was only going to buy what was available in the store (no 10,000 or 15,000 RPM drives to choose from), I chose the SMALLEST sized drive with the largest cache. IOW, 32MB cache is the largest size. I chose a 750GB drive as it was the smallest drive that would allow that sized cache. I will never fill the 750GB so maximum capacity was NOT an issue for me. If I could have chosen, I easily would have gone with a 10K RPM drive as 10,000/7200 = 25% faster for EVER and EVER.

Now, on to our discussion:

Let's exaggerate. You have a tiny 100GB 7200 RPM drive. On the outer part platters, there is a higher DENSITY of sectors so the transfer rates are greater on the outer part. This is where data starts getting stored first.

Internal Media Transfer Rate

Once you start fragementing on this little 100GB drive, you're already getting to the inner parts of the disc where transfer rates ARE slower. Much slower. Defragging this tiny drive will have a greater benefit than the drive discussed next.

Now we have a 1TB hard drive. Since this drive has so much space, getting to the inner part will take 10x longer than with the 100GB drive above.

So, whether you partition the 100GB into a 50GB OS adn 50GB data partition or take the 1TB drive and partition it 50GB OS and 950GB data, the 50GB will reside on the outer (fast) portion of the disc.

 

[PeteC]

fseal said it well: "Yeah the entire defrag thing is so... nebulous" AMEN!

Although all drives fragment over time, the typical user will never suffer enough disk fragmentation to notice any perceptible slow down. Windows Vista and Windows 7 do just fine handling the defragging chores in the background, and except for pretty extraordinary circumstances, the default settings are just fine.

There has been a cottage industry catering to the meticulous/paranoid extolling the virtue of constant defragging and third-party defraggers promisng enhanced system performance...however, the reality is that I have seen computers that ran for years without a single defragmation, that were not even close to 10% fragmented! I think that a good OS SHOULD handle this very mundane "house-keeping" without requiring user-intervention...

To answer the OP: 1% fragmented? Beneath even a minute's worth of bother...wouldn't even think of defragging over that!

 

[smarteyeball]

Prerequisite: I can't find the specs for small vs. large drives to see how many PLATTERS (disc surfaces) are involved. There are performance implications on how many heads vs. # disc surfaces, etc for a given size GB footprint.

(sic)

So, whether you partition the 100GB into a 50GB OS adn 50GB data partition or take the 1TB drive and partition it 50GB OS and 950GB data, the 50GB will reside on the outer (fast) portion of the disc.

Thanks very much the info Jim. Much appreciated.

 

[JimLewandowski]

PeteC is 100% correct.

Defragging isn't the issue. Keeping CLOSELY accessed/related files CLOSE to each other is. So, for a defrag to be useful, it would have to have access to the patterns and quantity of accesses for files on the drive (I'm talking non-data files).

Oh, wait, W7 does all that rearranging in the background.

I will never ever defrag my OS partition unless I need to because of lack of contiguous space (for other things).

 

[GordonLM]

Defragmentation harms SS hard drive?

I was warned by Defraggler that I had an SSHD and that defragmentation could wear the drive. I do not know if reading fragmented files is more wearing than rewriting them to the unfragmented state on an SSHD.

 

[Lady Fitzgerald]

While the rule of thumb for SSDs has been to never defrag them to avoid excessive writes, over time, reports have been popping up saying they do need an occasional defrag (but nowhere near as much as a HDD). I have yet to defrag any of my SSDs because the fragmentation level is way too low to worry about or, in the case of the boot drive in my desktop rig, the fragmentation level stays at 11%, which is not enough to worry about as long as it stays there (I've been running pretty much 24/7 for almost two years).

As far as defragging HDDs on Win 7 goes, I've found that Win 7's native defragger works as well and usually faster and better than third party defraggers. In fact, one of the ones that was popular with XP, Defraggler, was considerably slower and managed to corrupt one of my HDDs beyond repair other than a reformat (thank God for backups!). I have Win 7's native defragger set to defrag as necessary during the wee hours every Wednesday morning on my desktop computer (my notebooks only have SSDs for system and storage) and, when I was still bothering to spot check on Tuesday, I never saw a HDD go over 1%; almost all of the time, they were still 0%.