To defrag or not?

This O&O tool estimates how much a defrag would help performance -- before doing it.

http://www.oo-software.com/en/products/speedcheck

(My result: 136.2& possible increase -- I only defrag with Windows, from time to time, and most of the time I use Returnil, which supposedly contributes to reduce the fragmentation.)

 

Eh, Windows says I'm 1% fragmented so I don't really need a second opinion.

 

Thanks for the link!

 

I defrag about once every 10 years.......

 

Looks like a pretty worthless utility they are using to try to sell their defrag product with empty promises of performance increases you will never get.

 

Reminds me of auslogic. "You have 13 registry errors! Critical performance hit!" lol

 

Couldn't agree more

Or the antivirus malware trojan/rootkits

 

The first time I saw Auslogics run a scan and report to me how horribly my computer must be running I actually thought "This is less convincing than most malware."

 

When I set up my new WIN7 computer last May, I installed my PerfectDisk defragment app too because I paid for it and had it on my old computer. After a few weeks of installing stuff and migrating files, and tweeking the new system it started to get a little unstable so I did a defrag. Since then it's been running great and is 1% fragmented after 4 months. I trust it will be a very long time before I need another defrag.

 

I defrag once in a while, generally when doing my system backup and maintenance - which in and of itself happens when it happens.

I like the philosophy pitched by Disktrix, though their claim of 300+ percent performance gain (like everything else on the internet) is greatly exaggerated. And on SSD's it's a non-issue.

I defrag, backup, and clean garbage (a measurable and sizeable amount) as part of my regular maintenance. The specific routine I follow has served me well for many years.

 

Years ago I used a download manager that was supposed to speed up downloading by 300%. Of course when you measured the speed gain it was 0%. 0% is about right for defragging apps too.

 

Is it really worthless to defrag? Why would MS include it in their O/S' with a scheduling option as well?

 

There was a time when defragging made a difference. Especially with the band positioning mechanisms of the early hard disks and their piss-poor access times. Every millisecond counted.

With modern hardware, a lot of the latencies and performance bottlenecks are hidden and glossed over with layers of cache and well-performing hardware. Load times are masked by other system activities and distractions. And as SSD's become more standard, defragging programs will fall by the wayside. An SSD can give you the same read times on a file regardless if it is in 1 fragment or 5000 fragments. No kidding.

I would say the real gain of a defrag job on a mechanical disk comes from making the head move less distance. You really don't want the head buzzing back and forth between the inner and outer edges of the drive.

So in that respect. Pushing all your commonly used files to one area can, but not necessarily, result in a barely measurable 5 percent performance boost. This can also be accomplished with partitioning.



All defragging programs *might* get you 5 or 10 percent performance gains, perhaps..
But 300% gains? What you smoking? Gimme Gimme!!

 

It'll be 300% in some areas but like... 300% of some ridiculously small figure. Like a >.1ms increase.

Still, if a drive were entirely fragmented like 60%+ you could see some benefits. I have no clue how you'd manage to fragment that much (perhaps over years) but I suppose disk access times might be sufficiently increased to the point where you'd see a significant performance hit.

 

Here is a real-world example of using DiskTrix' Ultimate Defrag.

When I first used the ud3 it was on a drive that hadn't been defragged all year long, or longer. Ok. Now. There's a simulation I like to mess around with and it loads about 1GB or so worth of code and graphical textures. Good enough.

On an un-optimized drive with the files all over the place and fragmented I got 14.5 seconds load time.

After running ud3 to place the files at the "beginning" or outer edge of the disk I re-ran the test. This time I scored 13.9 seconds load time. I gained about 1/2 second off my load time.

Be aware that the outer edge of the platter, irregardless of what DiskTrix says, is *not* the fastest part of the drive.

And by placing the files (in the same order as I had them on the outside edge) just a little closer in, at about the 20% mark so to speak, I achieved 13.6 second load times.

You may read about why this is so right here -- http://hddscan.com/doc/HDD_Tracks_and_Zones.html

So ultimately I saved 1 second loading time by spending several hours & a lot of wear and tear testing and optimizing and coaxing as much performance out of the hardware as possible. Can you believe it?!?!?!?!!! I saved me a whole 1 second on loading! YeeeeHHHAAAWWW!

I ran each test 10 times on 10 fresh boots, to help eliminate the effects of os caching and stuff like that. I also allowed the system to "settle down" for a moment after I got the windows pointer available.

A more real-world test I did was loading Office Word and Office Excel. The applications pop open in about a second. And after optimizing heavily with ud3, they *STILL* took about 1 second to pop open. What'd I do wrong here?

Now --
Another test I did involved making a directory with a thousand files, and putting 500 of those files at the OD of the disk, and the other 500 at the ID. Then I made a batch file to copy them in sequence to an SSD. This caused a butterfly action on the servo arm as the drive read one file then another and so on. One at the edge, and one at the inside. Back and forth, back and forth, till all 1,000 files got transfered. It took 1 minute and 8 seconds to complete.

Next I put all the files near the edge of the disk, and in sequence. This eliminated all the head thrashing and the only noteworthy latency was rotational latency. This reduced the test time to 41 seconds. The drive also made very little audible clicking because the head didn't need to jump all over town.

This, folks, is a measurable and worthwhile gain. But also be aware that it is a comparison of an artificially created bad situation against an ideal layout. A real-world drive won't have files ordered like that to begin with.

My personal conclusion, a modern system won't benefit much from a thorough defrag unless it's bad shape to begin with.

 

Thank you Keatah.

Though I know I had a computer without defragging for about a year and it was barely fragmented. I would have been curious to see the actual fragmentation for a clearer result.

If it were a 5% defrag and it lost you 1/5 a second frankly I'd say defrag anytime you're at 10%. If it were a 50% frag'd drive idk if I'd bother.

Cool link btw, good to know about short stroking.

 

This is an update to my post. I just bought Rollback Rx and it states that if you want to do a defrag you first have to uninstall Rollback Rx. So before installing Rollback Rx, even though my HD was only 1.4 percent fragmented, I decided to defrag it so I wouldn't need to do it again for a looong time. I started my PerfectDisk, clicked SMARTPLACEMENT to start the defrag, and the progress bar started doing an Analyze procedure. The progress bar got almost half way across the screen and just wouldn't go any further while it kept analyzing the same spot over and over it seemed. I let it go for 2 hours figuring that it is taking so long because I have a 1.5 TB HD. Well I finally stopped it and it was as if nothing had happened. Then my computer started screwing up with my firewall making MalwareBytes put up warning pop ups that a file in my firewall was a trojan. I quarantined it, but when I checked in quarantine there was nothing there. Well to make a long story short, I straightened everything out, but don't know what happened in my PerfectDisk. Now I'm afraid to use it. Does anyone have a clue as to what happened?

 

Thanks Keatah, nice feedback

 

I once defragged and it got interrupted by overheating. The whole computer started acting really weird. Lots of stuttering.

Defragging uninterrupted fixed it.

 

Hungary Man, Thanks for the response. I don't know what it was, but your experience could have been a possibility.

 

My view these days is that a periodic defrag that puts files in sequential order might not do much for speed, but does keep head seeks to a minimum. While head seeks might cost you a second of time, IMO it also introduces a quantitive amount of motor action that is not needed.

Head starts at sector 1, then skips to sector 8, then 24, then 17, then 20, then 9, etc etc to find the parts of the file(s). This is a fragmented drive.

Head starts at sector 1, proceeds sequentially to sector 4 to find parts of the file(s). This is a non-fragmented drive.

On a data drive that has hundreds of gigabytes of data, the amount the head travels and the times it travels can easily be heard on a fragmented drive. Defragging it is a perceptible sign that the head is doing less seeking, as your ear can plainly hear the difference.

Whether excessive seeking can degrade a bearing etc over time is anyones guess. Whether sequential or date or even access fragmentation schemes are best, again, anyones guess. All I do know is that I can hear the difference, and from a laymans standpoint, defragging my data drives a couple times a year (if I remember) certainly doesn't hurt and might help.

Sul.

 

Good points you bring up, Sul! We all were looking at it from a speed perspective, but you wisely point out the stress placed on the head from all the seeks it has to do

 

I think everyone can find a lot of useful information in manual to UltimateDefrag (Disk Trix)...how disk works, how defragmentation decrease speed and performance of disk, ect. It's great for me.
http://www.disktrix.com/ud_userguide.htm