Defrag opinions

I'm testing the new version now.

When I installed the programs I was told to open a port on the Windows firewall. When I uninstalled the uninstall program didn't remind me about the open port, so it remained open until I remembered to close it. I think it had something to do with copy protection.

I don't think version 1.72 requires any firewall changes. It hasn't asked for any.

This is easier said than done. I'm not sure whether this kind of experiment is repeatable. I'll tell you when I've written my program to create fragmented files. As I said before, the number of new fragments is not necessarily an issue, but the file performance associated with those fragmented files I.e. where the frgaments are placed).

Certainly no defrag vendor has published ANY figures about the performance of their programs. DK claims % figures, but never explain what the % represents. I can claim a 10% performance improvement, but it doesn't mean anything unless I say what it is a percentage of.

 

Donn,

Not easy at all, infact when I tried, I used about 50 files and minimal difference between products, need to think bigger scale.
Need to simulate an average days use on a PC. With a mixture of files being modified (simulate log files and documents) and files being deleted and created (simulate temp files and browser caches, the odd attachment from email).
I would think we would need to look at thousands, possible even hundreds of thousands of file operations to get meaninful results. As you say, very hard, especially to make it repeatable.
Improvements could be measured against not defragging at all, using fragments as units.

 

Using Diskeeper 10 and Perfectdisk I ran a test to measure refragmentation levels. Both using 100% default settings.
Hdd is a 20gig virtual drive with 23% free space.

There was minimal difference between diskeeper and perfectdisk as to refragmentation level. I ran the test 3 times for each defragger to confirm and the results were very similar each time.

So what did I do ?

Reverted to previous project version in visual studio (project under version control), and my test text, zip and sql files replaced with originals.
Cleared Browser cache. Close Browser and all running apps.
Clear windows/temp and user profile temp.
Then the drive was scrambled using Raxco's scramble tool http://www.raxco.com/support/kb_detail.cfm?action=options&prod=1&ver=8.0
The drive was then defragged.
Diskeeper left a few fragments so I re-defragged and the remaining file was defragged (using perfectdisk to confirm).

I Counted the number of files and number of fragments (using analysis of both diskeeper and perfectdisk, they reported different file counts, but gave the same number of new files created and also the same fragments)
I then rebooted.

I then did the following, which represents a sample of what I do during the day...

Browsed to Google, Wilders, Ebay and then Gmail.
Added specific text to a Test text file.
Updated a test zip file.
Updated Visual Studio project from source control
Edit SQL file in SQL Manager
Close all running apps in task bar.
Count files/fragments.

This test was repeated using Diskeeper and Perfectdisk 3 times, the number of files and fragments created/modified only varied by 1 at most.

The (averaged) results...

Diskeeper 230, new files were created, 293 fragments created
Perfectdisk 230 new files again, 298 fragments created.
Dont forget that quite a few files were modified as well.

These results are very similar - the difference is insignificant.

I will not draw any conclusions, but thought I would share, just to show how hard is it to measure refragmentation.

One thought I have had is that there is something more significant

To be scientific, the test needs model a week or 2 weeks worth of simulated usage, then we might be able to see any significant differences.

 

Nick can you try with jkdefrag?

 

This is intriguing. Please could you measure the read times of the new files using my current version of Prefetch

http://www.openaccess.co.za/BlackAndWhiteInc/PrefetchSetup.exe

You'll need to create the list of files to be tested for file performance by hand, because I doubt if the prefetch.ini file will have those new files in them.

Thanks for the info on the PD utility to create fragments, and the file access timer. I had no idea they existed.

 

Hi Donn.

I will have a go.

FYI, an easy way to list files to a file in command line is to cd to the directory where you want store the file, the run "dir /s /b > yourfile.txt" and the output will be piped to a text file

 

OK. The results are in. Again, using DK 10 and PD 8.

I ran a slightly different test this time, including timings.

Again this was based on something that I did today.


Cleaned temp folders.
Rolled back to a previous project in visual studio (under source control).
Restore backup of files in my test folder.
Use Raxco's scamble to fragment my drive.
Defrag.
Reboot.
Edit a php file.
Edit an SQL script.
Get latest (including a replace) of my project in visual studio.
Update a zip file with the project (over 60 mb).
Make a copy of old project backup folder and then make a copy of existing project into one of the folders.

I had to do a test run first, so that I could then get a listing of the files used in this test, so I could then benchmark them.

The results...

Diskeeper.

Before the reboot
79155 files reported by Diskeeper.
90538 file reported by Perfectdisk.
Both report no fragments.

After the reboot and test run.

Diskeeper reports 80834 files and 1086 fragments.
Perfect disk reports
92328 files, 1039 fragments.

2341 files were used in the timings, using Donns tool.

1.00 seconds, average of 3 runs. (Note that I didnt use the first run as always higher, everytime, I assume caching coming into it).

After defragging again (0 fragments reported only single pass needed with DK 10)

The read time dropped to 0.95 seconds, average of 3 run.

Perfectdisk

Perfectdisk reported 91013 files
Diskeeper 79625 files, 0 fragments after the intial defrag with Perfectdisk.

After the reboot and test

Perfectdisk reported 92705 and 1001 fragments.
Diskeeper reported 81270 files and 1025 fragments.

Timings

1.08 seconds, I forgot to put this in my notes that I emailled myself from work , but I am pretty sure the time dropped to 0.97 seconds after defragging with diskeeper.

So it appears that perfectdisk reduced refragmentation by approx 4% compared to diskeeper, but why was the performance drop about 7% greater ?

Well, This screenshot might explain:

http://img98.imageshack.us/img98/2882/pdkc0.png

There was also a chunk of files near the end of the block files which were fragmented (dark red). As you can see for whatever reason a big block of newly fragmented/created files were placed right at the end of the disk.

In diskeeper this did not happen (free space was in about 6 chunks).

Again, im not drawing any conclusions.
But I still feel the number of files used is still too low so I would not put any significance on the data I collected.

 

First time i tried PD this summer i realized spectacular results but soon turned to dismay when it began to exhibit signs of deteriorating performance, not the disc but the software itself. I quickly dumped it for Ultimate Defrag and have never looked back. UD seems intent on rapidly building more and more improvements into it in record time, some times weeks between upgrades.
I know i lost cash on Diskeeper though, it proclaims a lot in PR about Invisi-Tasking and other enhancements but affected some performance in other areas and i simply don't trust a defragger running in the background all the time although thats suppose to limit fragmentation and boost percentages of performance. It done nothing of the such for my machine but Ultimate Defrag makes strategic placement "EXACTLY" where i need them to be and puts the user in control, not a software. I enjoy the interaction of setting up the order of the layouts, and then UD carries out those orders to perfection, and the results are unmistakable plus remarkable.

My Opinion

 

Could you explain ? how does the software deteriorate or show signs of deteriorating ?

 

The port doesn't remain "open". If there's no program named UDefrag.exe listening on that port then it will remain closed.

 

i know something about ISRService.exe,it also has a port open to communicate,in my case i figure it is port 48288.but for whatever reason they need it,i suppose it does't hurt. By now they dropped all support for these goodies so verifying or whatelse is a deadend,

 

Can anyone try JkDefrag and tell me if they like it or how good it cleaned?

 

I switched from O&O to JkDefrag just recently. No services, no extra processes. Very light, fairly quick, does a good job.

 

I forgot to mention, you can guess what the before screenshot for perfectdisk.
Here are the before and after screenshots for diskeeper (using perfectdisk analysis).#

After initial diskeeper defrag
http://img123.imageshack.us/my.php?image=dk1ko8.png

After the fragmentation run
http://img509.imageshack.us/my.php?image=dk2vi6.png

 

nick please test jkdefrag . Wanna see if this neat freebie can whoop commercial defraggers .

 

JKDefrag - I will see if I can test it monday (default settings only). Only potential issue is that it does not respect the layout.ini prefetch file of XP.

 

There is a discussion about it. Maybe you would like to read it.
http://www.kessels.com/forum/index.php?topic=319.0

 

any new info from the new version of disktrix?

 

Where is the evidence (other than that provided by the firms selling defraggers) that defragging has any effect on computer/disk speed? I have not found one piece of objective evidence that defragging has any measurable impact on performance.

I own Perfect Disk and Diskeeper and have measured before and after defragging performance- and there was no measure of performance enhancement- none.

 

just think does any other operating system need defragging?
or is it just that microsoft has a bad inbulti defrager?
if the answer is no then if microsoft made there fileling system bettter than home users and big corporations wouldnt need to waste there £££ or $$$ on expensive defragging software.
lodore

 

The degree of improvement resulting from defragging really depends on how badly your system is fragmented. The last independent study I know of was done with WinNT/2000. But since WinXP uses the same kernal, the conclusions should be applicable. Here is the white paper.

 

I agree - people do talk a load of BS about the benefits of defrag programs - especially ultimate defrag - but the reality is that without a defrag program performance would suffer over time. This does not mean that files need to be moved every 5 minutes or rearranged in some ridiculous way. Any basic program will do - even the built in Xp program - and can be run perhaps once a month.

 

It really depends on how much your computer, your mileage may vary.

 

I have yet to see any evidence other than contentions (assertions) that performance is enhanced. Saying you think it is, does not make it so. Where's the proof?

 

Why not comment on the above white paper? ..