Multi-pass mess

Long ago, Gartner concluded and reported that "a multi-pass defragmenter must be run over and over to defragment the disk, especially when defragmenting very large disks with heavy defragmentation and limited free space. As such, multi-pass defragmenters characteristically fragment the remaining free space on the disk, which accelerates fragmentation later.It is recommended that a third-party single-pass defragmentation tool be implemented instead."

This is one reason PerfectDisk employs a single-pass approach. Besides being Microsoft's only certified defragmenter, one of the things that makes PerfectDisk unique is its patented single-pass SMARTPlacement technology and Space Restoration Technology. These technologies not only speed up boot times and future defrag passes, but also consolidate free space into a large contiguous chunk to proactively cut the speed of refragmentation by up to 90%. Multi-pass defraggers, that can't be scheduled, only react to fragmentation in real time by attempting to defrag after fragmentation has already occurred - all while taking up valuable system resources when you need them most.

Despite the findings of the world's leading information technology research and advisory company, some - or actually one - move further and further away from the strategy recommended by Gartner. That one issues press releases that state, for example, "many have discovered that fragmentation continues to build between scheduled runs. These large drives require a more modern approach and specialized solutions for large-capacity drives. Scheduled defragmentation, which leaves fragmentation behind between runs, is outdated; transparent, automatic defragmentation which runs whenever unused resources are available is the order of the day."

Uh...no. If a multi-pass defragger runs over and over and fails to adequately consolidate free space, then yes - fragmentation would build rapidly. But that's why Gartner recommends not doing that!

Uh...no. A specialized solution for large-capacity drives is not modern, but a rip off. That's why, when PerfectDisk was architected to support multi-terabyte drives before any other solution could come close to handling them, we determined that the solution would work for all drives, and there was no need to charge extra for it. So we don't.

The "order of the day" is still single-pass, despite wild and unsubstantiated claims from a lone cry in the wilderness. Clean. Efficient. Fast.

No mess.

               

Comments

[PerfectDisk]

Hi Agreebutdisagree,

Thanks for your comments. I understand argumentum ad verecundiam, but was only using Gartner as an example of a reputable source that came to this conclusion. It's similar to the same logic why we, or any vendor, makes note of recognized companies or government agencies that make a substantial purchase from us. So, I don't expect the fact that, for example, UBS has purchased 90,000 copies of PerfectDisk for their corporation will mean you will rush out and buy PerfectDisk. That type of news is meant to convey a level of trust and due diligence that a company with UBS' resources has given to reach their decision. And that is why I always say that whatever white papers or information we talk about should be used only as a basis for one's own decision.

Regarding the Gartner report, we don't and aren't authorized to publish that. You would need to obtain from them.

Regarding the overall philosophy of multi-pass versus single-pass, here is the general concept...PerfectDisk defragments the drive as thoroughly as possible, with a strong emphasis on free space consolidation. With that free space consolidation, new files are much more likely to be created contiguously. With the multi-pass strategy, because there is only an attempt to chip away at the fragmented files and free space, as new files are created, there typically are only small fragments of free space available, so new files are fragmented from the get-go. We could create any scenario that shows this, but that wouldn't be a real-world scenario, it would be coming from a vendor, and you would be rightly suspicious.

The best scenario is to do your own testing on your own systems, so you can see it actually take place. If this is for a corporate purchase, I would also encourage you to schedule a demo with your account rep.

Thanks,
Joe

[agreebutdisagree]

Me again. Could you please publish some of the tests on slowed refrag you did internally? I understand the conditions, but proof of concept helps for purchase decisions.

[agreebutdisagree]

As I am trialing Perfectdisk for a corporate license I wanted to read this paper but could not find it anywhere. Can talk to my sales rep to get it? Based on the limited published quotes, I agree with the technical findings that the built-in defragger is crap, but not the illogical conclusions. Before changing careers I was in business law. What Gartner has done here is a fallacy of logic; a non-sequitur conclusion.

It's also a fallacy to point to the word of an authority that has applied false logic. It is called argumentum ad verecundiam.

Maybe I'm wrong, as I'm basing these comments on the limited information published.

[PerfectDisk]

Hi Wait,

Thanks for the comment, appreciate the input.

Regarding the use of the term "multi-pass," that is "their" term -- they claim to have a patent on it, although we never see a patent number and don't know why they'd want one on it. And it's also the term Gartner used to describe what they do. In any event, you raise a good point about it shouldn't necessarily be "our way is the only way," so we're adding a new way closely aligned with that other way. However, we still believe the single-pass approach is the best. And with consolidated free space, there is less fragmentation and slower refragmentation. One issue is that is a user is doing all those things you mention, than the computer is not "idle" during those times, so the defragger won't do anything (according to that "other" way). We try to address some of the issues you mention with the single file defrag, which has proved widely used, and we are expanding its capabilities.

So if you don't like schedules, you'll have even more to your liking with PerfectDisk in the near future.

It's not our way or the highway. We strongly believe in our overall approach, and Gartner agrees. But we try not to be so close minded that if you want another way, you'll have it.

Thanks again - I really do appreciate your comments.

Joe

[Wait...]

Why are these two approaches discussed as if they were mutually exclusive?

I've been wishing for years that PerfectDisk would employ a real-time (or as you call it, not quite accurately, "multi-pass") approach, along with its current feature set. There is no reason that a defrag utility couldn't detect and defragment files in a real-time sort of way, and then also do a more thorough space-consolidating defrag on a regularly-scheduled basis. Or, the product could use only real-time on some disks, and only scheduled defrags on other disks.

I know, the claim will be that the real-time defragging done on the files is a waste, because they're just going to be moved again later by a scheduled defrag. My reply to that is, as you like to say, "Uh...no." It's not a waste because although free space won't be consolidated immediately, at least the files aren't badly fragmented. This matters to me when I have large files that are in hundreds of fragments, and want them defragged *now*! And again, on some disks, it's not only sufficient to use only a real-time approach, it's essentially mandatory. So that scheduled run may never occur at all.

I know the rejoinder to this, too: With PerfectDisk and scheduled defrags, those large files won't be so badly fragmented. Again, "Uh...no." I have some volumes that aren't accessible all the time (network drives, encrypted volumes that come and go, etc.). I can't really defragment these drives on a schedule.

And you know what? I hate running schedules even on my primary drives. You say that the "multi-pass" approach uses too many resources (that's not true, if the product is well-designed), but during the times you think I should be running a schedule with PerfectDisk, I have more important things to do with my disks, such as scanning for malware, running backups, securely overwriting free space, and so on.

I strongly feel that the "Our way is the ONLY way!" approach is arrogant and misguided. Both approaches have their advantages, and both can be used together--either within the same system or even on the same disk(s)--to get the best of both worlds.

[PerfectDisk]

Hi Ray,

Thanks for your comment. Beacuase we place a high emphasis on consolidating free space (to improve write performance and slow down refragmentation), we pack files together, and what you're seeinng could be from bad clusters, or files that are unmovable such as the pagefile or hibernate file.

One interesting thing for you to try (if you're really interested:) ) is to perform a "Defrag Only" rather than a SMARTPlacement defrag. In this case, PerfectDisk won't worry about consolidating free space so you could see fewer fragments, but you will also see more fragmented free space. But I can also say that the very few fragments you have remaining are not causing performance problems.

Thanks,
Joe

[Ray]

Speaking of multi-pass, I find that PerfectDisk never completely defrags my drives. It will always require me to run it several times, until I either give up or it actually achieves 100% defragmentation. I usually give up.

Don't get me wrong, I'm not trying to sound negative. This product is a much better solution than others I have tried. I just would like to know why it can't get all the files defragged in one pass.

Case in point: My 120GB disk is about 1/3 full. I ran the offline defrag first, then 3 passes of SMARTplacement. I now have 6 files left undefragged after the last pass. The largest of them is 6.7 MB with 4 fragments. The smallest is 0.1 MB with 5 fragments (!).

What makes the app decide that this is acceptable?

[PerfectDisk]

Hi Nick,

Thanks for your comment. It all depends on what the activity on your drive is. We regularly do internal testing,, and we have results that typically range from 40% to 90%. We test on drives from 20GB to multi-terabytes. I think when Gartner did their research they were in the 40% to 70% range. But I could manipulate number to show anything - even 100%. It goes back to each individual, and that's why I always say don't take our word for it. But it's a basis for understanding the methods and your own evaluations.

Thanks,
Joe

[Nick]

Hi. Do you have any figures to back up the claim of Smartplacement "proactively cut the speed of refragmentation by up to 90%" ?

Cheers Nick.