Defrag opinions

Donn,

I have been been reading through bits and bats, some good info there, though I gotta say I've never experienced your issues that you have had (not been in the same situation), so can't disagree with your conclusions.

What I can say is I've seen repeated times different defraggers work great when another's doesn't for different people.

Have you considered running benchmarks and/or fragmentation counts with respect to time, as one thing I would like to see how quickly the drive re-fragments and performance drops between defrag runs as this IMHO is more important than initial after defrag performance as a simple reboot boot starts to undo the good work. Also important is how longer further defrag runs take.

The results are interesting; they suggests that file placement strategy helps little with general performance. It also supports the fact the drive access is pretty much random (in general usage). What can help is reducing head seek times, which is what I suspect one of the things IFAAST trys to achieve (by grouping recently used (read and written) files together on the fastest part of the drive, but I've never saw any gains when I trialled worth the extra cost.
Also one thing to note is a file placement strategy could conflict with the way the NTFS driver writes data to disc and cause unnecessary re-fragmentation rates and increase in future defrag runs.

IMHO the approach of optimising file position to reduce future fragmentation is important and this is one thing that many Linux file systems do at a driver level when writing data, which is why they are far more fragmentation resistant and do not loose performance with age (which is what the problem with fragmentation is). Diskeeper's multi-pass approach and perfectdisk's Smartplacement both try to achieve this in different ways.

If I had the spare time to sit down and measure re-fragmentation levels and performance levels with respect to time (and subsequent defrag runs), we might be able to draw even better conclusions.

Keep up the good work, Nick.

 

In regards to an often seen reg tweak in disabling time stamps.

Could this have an adverse effect on say PD and using the default 60 and 30 days settings for smart placement defrags?

 

I don't know with PD but with UD you have to enabled it.But guess with PD is the same.

 

Thanks Huupi, yep I'd take a guess and say no to disabling time stamps as well.

 

The test system does not have time stamps disabled, because I used the default WinXP settings as much as possible.

PD8 doesn't use "Last Accessed Time" but rather "Last Modified Time" as far as I can tell.

I agree. remember that most of the results shown are measured when there is NO fragmentation left. Only the "Basic XP" and "Basic Office" results include fragmented files.

I think my biggest conclusion so far is that the tests don't reflect user experience or usage. To that is going to take a lot more thought and work.

BTW, I am not ignoring UD from DiskTrix. It's on the list.

 

Donn Edwards, excellent job, its quite daunting to test out all the defraggers out there, in lieu of that you seem to have done a swell job. My own experience tells me that even though JK Defrag does an excellent job, for best boot times, its PD with its excellent offline defrag.

 

Interestingly I have my pagefile in 6 fragments on a work machine that will not defrag with O&O, mst defrag or pagedefrag, going to try diskeeper and perfectdisk tomorrow as not in the office today.

Its a 20 gig drive (vmware virtual drive) with 27% free space, page file is 1.5 gig approx, so should'nt be any issue with it defragging.

 

I use Windows XP's defrager.

It does a good job (sometimes it takes 2 or 3 times to get the files nice and tight).

I defrag in safe mode and it is really quick!! (a tip from an IT person).

Once I deleted a partition to the left of one XP and added this space to the XP partition which left the file in the middle of the defrag analyzer. It only took 3 times to "move" the partition using the native defrag utility (yes it had "moved" 2.3 gigs of OS completely from it's original position).

I thought this was impressive (no corruption, which even DD10 sometimes does when moving a partition) so I have confidence in it due to this performance.

 

I've tried PD before and realized some snappiness after it's defrag but that program is just not been compatible enough for my setup so i soon dropped it for UltimateDefrag by DiskTrix.

Now i do experience performance gains with UD but i also recognize some limitations compared to PD. I feel i threw money down the tubes when i recently picked up Diskeeper. An earlier version was unstable so i dismissed it easily but this latest one showed some promise early on, but i have a real problem with any defragger running in real-time and especially defragging TIF's and such that are to be discarded anyway. I'm sure some setting someplace on it might exclude that folder as i've excluded others but overall Diskeeper is proven a disappointment whereas UltimateDefrag is at least returned the results i expect plus it puts the user in control, and control of your own placements and such is key to satisfaction. Now it would surely be welcome if a defragger could auto-defrag to the user's exact wishes, but unfortunately, IMO, Diskeeper doesn't afford this freedom, UD does in spite of some limitations, none of which i might add are of real significance AFAIK.

So i conclude that UD (at least for me) continues proven, reliable results and passes CONTROL to how i expect file placement as well as features such as RECENCY, VOLATILE, etc. User interaction IMHO will always be paramount to optimum disk performance and extended life of the hard drive so long as it's applied practically and methodically even.

Diskeeper on the other hand i found maintains ongoing continous disk activity in the background and who's to say how far along an automated defragger of this fashion will eventually or soon tax the limits of these sensitive hard drive mechanisms to the point of contributing to early failure. That's my assumption anyway, others might have better insight as regards DK but from my experience with it and others, i view UD as my best safe choice for maintaining and keeping a healthy system with bare limit drive wear.

Please feel free to openly discuss and/or oppose my skepticism toward such an approach, but satisfactory results coupled with the satisfaction of keeping control steers me to stay with this choice of mine when it comes to these Super Defraggers as i like to call them.

Regards EASTER

 

With regard to any flavor of defraggers i think that differences are not that big in terms of layout,performance,at least with the well known Vendors.But what sets UD in my favour is their thorough helpfile [if you want to learn some more about disklayout and any related]and their commitment and dedication to their product.And yes i see a gain in diskperformance after defrag with it,saying that before i used Perfect Disk with is also a topgun.

 

I've actually suffered an issue with diskeeper's realtime defrag and that is not getting enough idle time to defrag the drive fully. It does defrag the odd file but not much.

BTW I agree with Easter on the subject of defragging temp files (and consolidating), it does seem wasted effort, if we could exclude recently modified and/or temp files (as they are most likely to be modified/deleted again) would be great, but then logically there would be little for a realtime defrag. IMHO it does no harm, but is pretty much futile.

Looking at the top 20 most fragmented files on my HDD they are a mixture of log files, a few app temp files, a few browser cache files and my user registry dat file. I last ran a defrag 2 days ago.
This also does support any file placement strategy that puts most recently modifed files closest to block(s) of free space.

 

Btw do you have windows/programs on 1 partition & data/mail/browsercache on a separate partition? Perhaps if they're not separated, then I can imagine a smartplacement feature can help more.

With mine on separate partitions, I find myself defragging my data partitions very infrequently. My windows/programs partition I'll defrag prior to saving an image, maybe every 2 weeks. In-between I sometimes do a restore (eg after trying out an app) - and it's something like a defrag too? So a basic free defragger like PowerD seems enough for my habits.

 

I have 2 machine at work one has 2 partitions and the other has 1.
By comparison the machine with 2 partitions, which is only a month old and not ever been defragged (previous list is machine with 1 partition) has a mix mostly system restore files, some printer dlls, and a few temp java files as the most fragmented files (1000 out of 35 000 files fragmented)
D: has a mix of temp files and email log files (total of 1000 out of 5000 files approx) fragmented.

 

Having played with Exec Software, windowsXp cut down version of exec software, Perfect Disk, O&O and UD I'm not really sure if all the effort is worth it.
The law of diminishing margin returns seems to operate with a vengeance when it comes to defrag. Use any defragmentation program on a system that has gone to seed and a noticeable improvement will be seen but I think that those who see differences in performance between any one of the major programs have very vivid imaginations.

My own solution is to defragment C: and then use a freeze program. At every reboot C: is as good as it needs to be. Trying to squeeze a little more out of the drive is a mugs game. Research to determine how many angles can stand on the head of a pin is quite a different matter.

 

Hi,

Having Diskeeper pro 2008, O&O Defrag Pro 10 and UD 1.72 on 3 different machines (mine and friend's), I am amazed by each one's unique features: DK and O&O have so-called background defrag powers plus off-line defrag. Whereas UD's offline feature is still a mystery to me, its flexibility and abundance of options could make advanced user a happy camper.

But in the end, what makes me very impressed is the Frag Shield available only in Diskeeper 2008 pro. According to them, this feature is to monitor MFTs and paging files to prevent MFTs from fragmenting and to offer optimal setting for paging files. How does it work ? This feature is apparently a very unique(?), perhaps some gurus can offer some insights ?

Take care.

 

I do defrag my data partition but not frequently/constantly. It might have x fragmented files out of Y... maybe it's just me but I can't really tell its effects, such that it'll make me think of more frequent defrags or smartplacement. Some files like newly created image files, I do prefer to un-fragment them (sometimes by moving them between partitions). But as for temp & log files...

EDIT: Reread my post, it's previous tone was soooo wrong

 

Frag Shield doesn't do much at all. If your drive is correctly organised your MFT will probably never fragment. Allow me to explain:

The Master File Table is a table of file names and sectors. Every time you create a file you add a record to this table. Once all the empty records are filled, the MFT grows. Windows (and most defraggers) keeps a section of your drive, (called the MFT reserved space) so that the new part of the MFT can be adjacent to the previous parts. In this way the MFT grows without fragmenting. The only way the MFT can fragment is if files are written in the reserved space, i.e. in the space reserved for the MFT, *and* the MFT needs to grow.

Frag shield tries to do this by making the MFT larger than it is at present. DK can also defrag the MFT at boot time, at least most of the time. Even if it doesn't, if the fragments are close to one another on the disk you probably won't notice any performance change.

What DK can't do is deal correctly with the other metadata files. The MFT is one of a number of such metadata files. PerfectDisk's boot defrag can fix up the other metadata files. I have noticed on a server that the metadata files grow and get fragmented pretty badly. I'm not sure if is connected with the number of users or with Exchange, but it can get out of hand fairly quickly.

Sadly I have learnt to take EVERYTHING that Diskeeper says with a lorry-load of salt. There is so much hype and misleading information that it is scary. Take the following statement as an example:

(emphasis mine)

I set up a drive with some of these "extreme" conditions and DK2008 made things worse. I seems that you *either* get file defragmentation *or* free space defragmentation, but not both, and certainly not under extreme conditions.

I filled up a drive with gazillions of fragmented files and only 2% free space, and DK2008 took days of automatic defragmentation and still got nowhere, distributing the free space all over the drive until it couldn't do any more. I have plenty of other examples of how DK breaks its promises, which is why I give them such a hard time.

 

I have been looking for some information on perfectdisks smartplacement strategy and the claims it reduces future fragmentation.
I have had a quick look through the white papers and not seen any figures to back their statement up.
Can anyone point me in the right direction ?

 

I'm not sure if any of the defrag companies have published figures about anything; it would certainly help me to work out how they measure stuff.

I'm also not sure how one could measure "future" fragmentation

The theory behind SmartPlacement is that by putting all the recent files in one place, along with the directories, any new files are likely to land up in close proximity, along with additional fragments of modified files. So even if a recently modified file grows and fragements, the extra fragment is close by, reducing disk seek times. And since the directory listings are close by, the hard drive is confining its seek range to a narrower band.

I think the best bet would be to post a comment on PerfectDiskBlog.com

 

I'm not pressing you, but curious about how do you rank UD in your upcoming review.

 

I'm busy running tests. I only started a few minutes ago, because of pressing work needs and a Christmas dinner last night. Give me a day or so.

You can read my product review of UD in the meantime.

 

Hi, Donn:

Thanks very much for your insights.

How is your views about DK's I-FAAST and O&O's automatic defrag ?

It seems to me that they both are very proud of these features.

Do they have a real value in term of performance ?

Hoping these features not slogans from then again.

Thanks, and take care.

 

I-FAAST costs $50 all by itself. Since DK doesn't do a "good" defrag in the first place, my feeling is that you are throwing good money after bad. When I did the benchmarks it claimed it would give me an "8%" performance improvment, and then said it only managed "3%", but if you look at the numbers the performance got worse. I have read the whitepaper and in theory it seems like a great idea, but the implementation sucks.

The automatic defrag in O&O conflicts with the screen saver, so you have to choose between the two. I think the JkDefrag screen saver works much better, and is worth checking out, even if you run it in -a 2 mode, which does a defrag but no special file placement, so it doesn't contradict any other defrag system.

The engineering on O&O is great, and they do a good job. My basic problem with automatic defrag is that it gets in the way if you're busy, and it could be busy on the file you decide to click on, which can be annoying.

If you like automatic defrag, then O&O is better than any of the other automatic defrags, including DK, simply because it works properly and does a proper job of defragmentation. It's the "set it and forget it" that DK advertises but doesn't deliver.

I prefer scheduled/manual defrag because you get to decide when it should run.

The only reason why I use PD8 and not O&O is the boot time and Metadata defrag in PD8 is brilliant. But I like O&O defrag a lot, and think it is much better value for money than DK, even though it is the 2nd most expensive defrag product on the market.

 

We could defrag, then run a set number of programs/operations to simulate some real use then measure how many new file fragments there are. Then repeat for a different defragger, we could even measure multiple defrag/usage runs.

I am having a look now, cheers.

 

Donn, I just read your review of UD v1.52. You might want to upgrade it to the current version (v1.72) as I think they fixed the large file issue that you alluded in your review.

I also have a question about your parting comment in that review...

I don't understand what you are saying. Would you please elaborate on that point?