Tuesday, January 27, 2009

PerfectDisk 10 Announced

"Raxco Software today announces Version 10 of the award-winning PerfectDisk family of disk defragmentation and optimization products. PerfectDisk 10 includes several enhancements and new features to help users keep their computers running at optimal performance, reduce costs and extend the life of their PCs, laptops and servers.
"Included in the launch is an exciting new virtually-aware product for businesses’ virtual environments -- PerfectDisk 10 Virtual Enterprise Edition (VEE). PerfectDisk 10 VEE is the first-ever centrally managed defragmentation solution that is both aware of its virtual existence and is also able to dynamically adjust its resource consumption behavior with respect to its physical server load, such as VMware ESX Server and Microsoft Hyper-V. Other defragmenters, if they can even run in a virtual environment, have no knowledge of their host environment, thus making them susceptible to performance-debilitating resource contention.
"A new PerfectDisk Home Edition is now available for home consumers.
"A complete list of enhancements can be found at www.perfectdisk.com/whats-new."

This statement arrived by email today. I will be testing the new version shortly, and a full review will be included in . They have provided me with registration codes for testing purposes. Home and Business users can find a comparison between PerfectDisk and Diskeeper on their web site. All the different versions of PerfectDisk for home users are listed here.
Update: Softpedia has published a review of PerfectDisk 10

9 comments:

Anonymous said...

i m using it 2 days now and i can't say i m impressed..it's the same old PD, plus a couple of new features(virtualization,ability to Pause and Resume) a revamped GUI and the same slow defragmentation.they claim 30% faster analyze and
defragmentation passes which is nothing but BS..for me it seems extremely slow like their previous versions.

Quant2 said...

From my experience, PD10 seems to analyse and defragment drives quicker than PD2008. However, 30% speed increase advertised by Raxco is huge exxageration, in my opinion.

To be honest, I was also quite dissapointed by PD. Since SmartPlacement is not so good for improving disk performance (it was designed to slow down refragmentation), I was hoping that PD 10 could at least offer some new defragmentation modes for those who prefer system performance over slow refragmentation rates.

Donn Edwards said...

"Improved performance: Saving you time and resources, PerfectDisk’s engine now runs up to 30% faster."

30% faster THAN WHAT?

Never trust a % sign that doesn't tell you what the two items are.

The 30% could refer to WDD, PD7, Windows 98, who knows? At first glance one would assume they meant the previous version, PD9, but it could refer to anything at all. I will ask for clarification, but it will probably take a few days to arrive.

Quant2 said...

Hello Donn!

Thanks for the reply. I didn't think about the fact that "30% faster" can refer to anything.

I would like to make a comment on PerfectDisk here. PD puts all the directories after frequently modified files, which means that directories often end up being far away from the begginning of hard drive.

Now, here is a description of how NTFS accesses a file:
http://files.diskeeper.com/pdf/HowNTFSreadsafile.pdf.

What this document says is that there is a lot of dialog between MFT and Directories in order to access a normal file.

Due to the fact MFT is typically located towards the beginning of a hard drive and it is common for PD to put directories far from the beginning of a hard drive, it follows that the usage of PD results in increased distance between MFT and all the directories. Therefore, when a file request is made, hard disk heads contantly have to travel a long distance between MFT and directories before accessing a file. This results in relatively slow file access times.

If PD put directories to the beginning of the hard drive, file access could impove as the distance between directories and MFT is much smaller in this case.

Donn, what do you think about this? Do you think that I am right about the importance of placing directories relatively close to MFT? Or I am missing something here?

Donn Edwards said...

Excellent question!

My experience of using PD is that the directory location and the MFT location is often not too far apart, resulting in little or no performance loss.

The placement issue was much more important in FAT systems because machines were slower and FAT worked differently to NTFS. Nowadays a lot of the directory and MFT information lives in memory cache, so the placement of the directories is not as important, and the fact that the directories are fragmented or not often has a bigger performance boost, because each fragment adds another seek time delay.

My gut reaction is that if Diskeeper says it's "A", it probably is only "a" or more likely "b". They have no grip on reality at all.

PerfectDisk does not force the MFT to remain at the beginning of the drive. Initially the MFT is near the start of the drive, as are most directories. As the drive fills up over time the location of both MFT and directories change, not only with PD, but with DK and most other defrag utilities.

The seek time to open *any* NTFS file is:
1. Seek time to Directory plus read time of Directory (plus additional seek times if Directory is fragmented); plus
2. Seek time to MFT plus read time of MFT (plus additional seek times if MFT is fragmented); plus
3. Seek time to locate first part of file, plus read time of file (plus additional seek times if file is fragmented)

Since the seek time to track 0 of the drive is not massively different from the seek time of the last track on the drive, the performance delay is caused by the number of EXTRA seek times, and less by the absolute position of the file. According to DiskTrix the difference is around 20%, which isn't massive.

Obviously things would be fastest if the drive could read Directory+MFT+file all in a single track or adjacent tracks, but this is essentially impractical, given the size of the MFT and the number of files on a modern hard drive.

I hope to show this in my FRAGG benchmarks.

Quant2 said...

Thanks a lot for your quick reply and for valuable information!

May I also ask you which defrag program(s) are you currently using on your PC?

Kind regards,
Giorgi

Donn Edwards said...

At present I have PuranDefrag 6 as my active defrag program on my Windows Vista Business laptop. I also use JkDefrag to diagnose the drive from time to time.

I am surprised that I have been able to "survive" since 28 October 2008 without installing PerfectDisk.

Anonymous said...

Hello again, Donn! I was unable to reply earlier because I was quite busy during the last few days.

For some reason, I am not able to post under my previous nickname Quant2, because I am getting error "URL contains illegal characters". So I have to post under Anonymous'.

It is nice to see you are using PuranDefrag as your main defrag program. I tried it and was quite impressed, so I will probably buy the program in the near future.

Regarding your previous post, I must admit that I overlooked the fact that MFT and Directories are highly cached which makes their relative position to each other less important.

I also agree with you that having MFT+Directory+File read in one go is impractical. But I do believe that at least MFT and Directories could be quite easily put close to each other (e.g. like UltimateDefrag's "Move Directories close to MFT" option.

However, regarding the position of the MFT, my experience is somewhat different from yours. I find that the position of MFT file on my system remains static despite the fact that my C:\ partition has been filling quickly in the past few months (too many TV Show downloads). Also, PD puts my MFT file at the same location every time I do a boot-time defrag.

In order to demonstrate behaviour of PD on my system, I am going to present some data here. I am using HDView tool in order to report exact statistics. My C:\ partition is 77.8 GB in size, or 20,408,565 clusters. PD places MFT file starting from cluster 786,432, and directories starting from cluster 14,234,702.

Now this means directories are 14,234,702 - 786,432 = 13,448,270 clusters away from the beginning of MFT. If directories were located in the beginning of the partition, this distance would just about 780,000 clusters. What this means is that if directories are located in the beginning of the partition (like after running JKDefrag), PD increases the distance between MFT and Directories by a factor of about 17 (13,448,270/780,000) on my system.

Of course, I am not saying that this will necessarily lead to a degraded performance, because MFT and Directories are havily cahed, as you said in your previous post.

Regards,
Irakli

P.S. I am really looking forward to seeing your FRAGG benchmark of various defrag programs soon!

Donn Edwards said...

@Irakli

The clusters may be numbered, but the time taken to reach ANY cluster is the same. This can apply to clusters next to each other because their physical location may be on different platters or tracks. On average, clusters very close to one another can be accesses faster, but not necessarily.

Also, clusters numbered with a difference of 50000 may be just as "close" because on the drive they may be in exactly the same position, but on a different drive platter. It all depends on the drive layout.

Until defrag programs start analysing the physical characteristics of a given drive and adopt an optimisation strategy accordingly, the positioning of the MFT, files and drectories (apart from the fact that they are contiguous) will not always speed up things except within a +-20% range.


Warning: The NSA and 4 million other sick weirdos with "security clearance" have intercepted this page and know that you are reading it.