Defrag opinions

Discussion in 'backup, imaging & disk mgmt' started by Huupi, Nov 22, 2007.

Thread Status:
Not open for further replies.
  1. Huupi

    Huupi Registered Member

    Joined:
    Sep 2, 2006
    Posts:
    2,024
  2. 19monty64

    19monty64 Registered Member

    Joined:
    Apr 10, 2006
    Posts:
    1,302
    Location:
    Nunya, BZ
    Wow, you got that right. No surprises as to the winners. Good to mention NTREGOPT + PageDefrag....
     
  3. yeow

    yeow Registered Member

    Joined:
    Dec 11, 2006
    Posts:
    225
    I have used PD8 in the past, and also tried JKDefrag recently (only for short while). Both added abt 10sec to my startup time compared to PowerDefragmenter. Got the 10sec off again after I defragged with PowerD.

    For normal operations like starting apps, opening files etc, I couldn't feel any difference when defragged with PD8 or PowerD.

    But PD8's smartplacement thingie took a longer time, esp when I deleted files near the "start" of the partition (then PD8 wants to "shift" every file after that). PowerD is usually very quick in subsequent runs, tho not pretty to look at (I minimize it anyway).

    I usually use PowerD in Safe Mode (can't confirm if that really helps), but in Safe Mode I have to minimize the Contig.exe window or else the scrolling text slows things down alotttt. I also use PageDefrag set for every boot.

    Of course, YMMV.
     
  4. NGRhodes

    NGRhodes Registered Member

    Joined:
    Jun 23, 2003
    Posts:
    2,331
    Location:
    West Yorkshire, UK
    I stopped reading after the 2nd paragraph.

    He compresses his SQL data and log files and wonders why he has problems... most knowledgable SQL Server DBAs know you shouldn't compress a database's data and log files.

    The problems he is having is not because of the defraggers, but because he is doing something that causes instability.

    http://technet.microsoft.com/en-us/library/ms191433.aspx
    http://msdn2.microsoft.com/en-us/library/ms179316.aspx
    And he wonders why defraggers are struggling with his file...

    http://www.sqlmag.com/Article/ArticleID/5122/sql_server_5122.html

    http://support.microsoft.com/kb/917047 (emphasis mine)

    His problem is not the defraggers...
     
    Last edited: Nov 22, 2007
  5. Huupi

    Huupi Registered Member

    Joined:
    Sep 2, 2006
    Posts:
    2,024


    I dont know much about server/corporate environment,but i think the conclusions about which is best defrag is no suprise,many people who write lovesongs on JK defrag are reconfirmed by this review.
     
  6. Arup

    Arup Guest

    Long time user of PD here, dumped PD8 for JK Defrag, some of PD's recent updates have been buggy, last one started probing for floppy drives and PD puts the PD service in background, JK Defrag is truly good but it needs a good offline defgragger, one that works in x64 XP as well.
     
  7. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    It's a pity you didn't try to find out WHY I made these choices.

    Firsly, as a software developer I have always been aware that the machine I work on is often more powerful than the machine used by the end user, and so I try to take this into account when developing software.

    Secondly, I only have 60GB on my laptop, so space is limited, and I really don't like to have to allocate 15% of this to some data files I don't always use.

    Given these two criteria, I decided to compress the files, knowing full well that this would slow down SQL Server 2000 considerably. The idea is simple: if the program can work under these conditions, it will work even better in a live environment. If it doesn't work on my development machine, then I need to rethink the solution until it does.

    I am fully aware that file compression is not good DBA practice, and would never think of doing this to a server or production environment. At least give me credit for that, even if you don't like anything else I said.

    The existence of these large compressed files only presents a problem to badly-engineered defrag programs, but it also separates the wheat from the chaff in a field where it is difficult to distinguish between a "good" program and a "bad" one. A good program like JkDefrag can handle large compressed files without too much effort. Even Contig can handle them without a problem if there is enough free space available to defrag the whole file. Why should I pay money for a program that can't?

    I have grappled with the problem of determining how to measure the startup time. I have noticed that the hard drive can remain busy for up to 2 minutes after the desktop appears. It's easy to click "start" on a stopwatch, but what event do you use to know when to click "stop"?

    Any suggestions would be welcome.

    I have tried the boot-time defrag option in UltraDefrag, which isn't as good as PD8's boot-time option, but seems to work ok in Windows Vista (PageDefrag doesn't). It's also configurable and has 64-bit versions.

    I am busy benchmarking all the packages reviewed, and would appreciate input and comment on the testing program and how to measure the speed of data files. So far I am only testing system and application files, which is only part of the picture.
     
  8. lodore

    lodore Registered Member

    Joined:
    Jun 22, 2006
    Posts:
    9,006
    hello DonnEdwards,
    how comes you always seem to give diskeeper such a hard time?
    thanks in advance
    lodore
     
  9. Huupi

    Huupi Registered Member

    Joined:
    Sep 2, 2006
    Posts:
    2,024
    From what Donn said its quite plausible that he has doing it his way and then there more knowledgeable people in there who discusses from their IT job perpective.
     
  10. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    I'm allergic to hype, and DK is IMHO totally overhyped, unless you think that it is OK for a $299 program to refuse to defrag a server drive. My recent tests of DK 2007 show that at best it is only 2.5% faster than the built-in Windows Disk Defragmenter. It refused to defrag my hard drive, so I uninstalled it. Here are more screen shots.

    The latest version, DK2008, has numerous bugs, and is only marginally faster than DK2008. I will post the numbers shortly. Yet the upgrade from DK2007 Pro Premier to DK2008 is $59.95, more than a new copy of PerfectDisk or O&O Defrag. It just doesn't make financial sense. I wrote to the PR of Diskeeper to ask what features can justify the $60 price tag, but got no reply.

    I have been writing software since 1981, and worked in numerous IT environments, including a large newspaper with hundreds of computers, both PC and Mac. At present I help to maintain a WAN that consists of 25 sites around the country, with both Windows 98 PCs and XP laptops. In understand about system performance, application deployment an ignorant users. I accept that others have a different perspective, and that my opinions are personal. I feel cheated by DK because their product doesn't do what it is supposed to do, and yet I paid good money for it.

    I hope that clarifies things a bit more.
     
  11. lodore

    lodore Registered Member

    Joined:
    Jun 22, 2006
    Posts:
    9,006
    so whats your saying is perfectdisk is both cheaper and does a better job of defragging?
    diskeeper did refuse to defrag the fat32 drive on my sisters laptop.
    i mean why did acer format it as fat32 in the first place for windows xp sp2 i dont know.
    i have got fast replys when i have emailed diskeeper in the past.
    in your tests which proved to be better perfect disk 8 or 0&0 defrag 10?
    lodore
     
    Last edited: Nov 22, 2007
  12. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    PD8 costs $39.99
    DK2008 Pro costs $49.95
    O&O Defrag costs $49.95
    JkDefrag is free

    Only PD8 has the ability to do a metadata defrag (part of boot defrag)

    Of the 4 programs mentioned, DK2008 has the worst defrag ability. I am currently benchmarking O&O Defrag, so I can't comment on speed at this stage, but I plan to benchmark PD8 soon, as well as Vopt, JkDefrag and UltraDefrag. So many programs, so little time!

    Acer does its install as FAT32 for reasons known only to them. It was easy enough to convert to NTFS and then tidy it up.

    I'm still trying to figure out how to benchmark defragging of data files, and what to measure. Any suggestions will be most welcome.

    FWIW, Steve Gibson, the inventor of Spinrite, also uses PD8, as well as Vopt, and Jerry Pournelle swears by Vopt.
     
  13. Huupi

    Huupi Registered Member

    Joined:
    Sep 2, 2006
    Posts:
    2,024
    If you already investigating into defraggers then its almost not possible to overlook ULtimate Defrag from DISKTRIX,cause it has an passionate following here on wilders and IMHO the best made defragger ever.Completely user configurable and a well thought genial concept,and best off all,it really makes a difference,performance wise.
     
  14. Long View

    Long View Registered Member

    Joined:
    Apr 30, 2004
    Posts:
    2,295
    Location:
    Cromwell Country
    FWIW I certainly find this all very interesting. I have been unable to get any benefit from UD so can not go down as a passionate supporter.

    @DonnEdwards. You praise PD8 and this I can understand - I use it on all of my machines. You also praise JKDefrag. Do you use both on the same machine ?
    Having just run JKDefrag and then run PD8 analysis I find that PD8 now reports
    that the different "colors" are all over the place. Confused as to how both can be good but place fragments in such a different way.

    All my machines run with either DeepFreeze6 or Returnil - so I get C: the way I want it and then defrag and freeze. will be interesting to see how JKDefrag compares to PD8.

    You say that disk still working for 2 mins. I can certainly confirm this. If I boot a program during this window it takes noticeably longer to boot then if I wait. I have minimised the number of programs services loading at startup but can still find no way to boot in less than 55 sec with the extra 2 mins after the desktop is operational
     
  15. NGRhodes

    NGRhodes Registered Member

    Joined:
    Jun 23, 2003
    Posts:
    2,331
    Location:
    West Yorkshire, UK
    Hi, Donn.

    Thanks for responding, I shall have a read in depth now I more insight and someone to answer my questions :D - I have seen so many people spreading FUD about software, especially defraggers, so tend to be very skeptical.

    I have written quite a few things on the subject of fragmentation on NTFS (and on ntfs file allocation) that might interest you. Have a search on this forum or I can point some threads out to you later.

    BTW the sector mapping issue affects SQL server 2005 in read/write mode as well as 2000 (in read/write and read only) for compressed ntfs - and it CAN conflict big time with diskeepers realtime defragging for sure (because the compressed data fragments so rapidly diskeeper cant keep up).

    I am pretty sure that DK2007 (cannot comment on DK2008 yet) struggles to defrag a file when there is not enough space to move it entirely to a temporary location and is a weakness.

    I know what you mean about testing on a old machine - great for spotting bad (inefficient) code... but in 10 years I've never experienced the level of fragmentation on a live DB or webserver that you have on your dev machine :D !

    Cheers, Nick.

    PS, I have 10 years professional experience as a developer (web and databases mostly) and admin related systems on the Windows platform (nt4 SQL 7 onwards) and currently one of the developers on a system to share patient records nationally and online (my bit).
     
  16. yeow

    yeow Registered Member

    Joined:
    Dec 11, 2006
    Posts:
    225
    Ohh, I didn't know abt the 2 minutes. In my examples, I used the simple stopwatch method - Start time when POST screen appears; Stop time when my XP's "Wireless Network Connection" tray icon loads & "flashes" (it's my last tray icon to load).

    Just now I dwnloaded & installed Microsoft's Bootvis utility, and let it run a "trace". I don't really understand what it does, or what the plots mean, but:
    - The Cpu Usage plot came down from 100% at around 53s, which corresponded to my stopwatch timing.
    - But I also do see some activity at 59s & 74s, perhaps that's the 2 minutes window you mentioned?
     

    Attached Files:

  17. markymoo

    markymoo Registered Member

    Joined:
    Sep 25, 2007
    Posts:
    1,212
    Location:
    England
    Perfect Disk 8 great program and thankfully theres the x64 version. That guy wants to try Raid it get's fragged alot more and faster.
     
  18. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    I use PD8's boot defrag about once every month or so, and only run a full defrag when the file organisation gets really bad. For the most part I have JkDefrag's screen saver to tidy up when I'm not busy.

    I realise these two programs have different approaches, so I try not to get them to "contradict" each other. I have tried running without PD8 altogether, but I can't. I end up "hauling out the big guns" if the file organisation gets out of hand.

    That seems perfectly logical, and thanks for the suggestion about Bootvis. It definitely gives me something to work with.

    I agree, and I don't think it's fair to expect the defrag program to keep a compressed file defragmented while it is in use. That's asking WAY too much. When defragmenting the data files I ensured that SQL Server wasn't running, so that the files weren't "locked".

    It has been great chatting with you guys.
    Security conscious people tend to think more about what they are doing. You may have heard my defrag shootout mentioned on the "Security Now" podcast with Steve Gibson. If you haven't listened yet, it's well worth starting at episode 1.

    http://bp1.blogger.com/_vCXBX4q7xEo/R0VVxjHs55I/AAAAAAAAAg4/BB3iNOQZF8U/s400/DK2008Chart.gif

    Here are the results for DK2008, but you'll need to read the blog post to see what the lines mean. After reading the enthusiastic messages on the DK web site, I really didn't think that I would end up paying $49.95 for a program that works 3% better than WDD.
     
    Last edited: Nov 25, 2007
  19. AaLF

    AaLF Registered Member

    Joined:
    Feb 20, 2005
    Posts:
    986
    Location:
    Sydney
    With PD8 the resident Defragger wouldn't using Microsoft's Bootvis utility be a situation of Bootviz tearing down PD8's good work and re-arranging the startup resources on the rim to its liking and then PD8 following after and tearing down Bootvis's good work and re-arranging the Start-up resources on the rim to its liking and then Bootviz following after and tearing down PD8's good work and re-arranging the startup resources on the rim to its liking and then PD8 following after and tearing down Bootvis's good work and re-arranging the Start-up resources on the rim to its liking and then Bootviz following after and tearing down PD8's good work and re-arranging the startup resources on the rim to its liking and then PD8 following after and tearing down Bootvis's good work and re-arranging the Start-up resources on the rim to its liking -more or less?
     
  20. yeow

    yeow Registered Member

    Joined:
    Dec 11, 2006
    Posts:
    225
    Don't "optimise" with Bootvis if u dun want it to mess with PD8's work. Just run a "trace".

    EDIT. Oh u're referring to XP's built-in optimising?
     
  21. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    How does the "trace" option store the information? If it's writing to the disk as the bootup is happening, will this interfere with the boot time?

    Perhaps we are dealing with an example of Heisenberg's Uncertainty Principle here: i.e. making measurements can change the results. Of course it would change them equally for all apps being tested, but still ...
     
  22. Franklin

    Franklin Registered Member

    Joined:
    May 12, 2005
    Posts:
    2,517
    Location:
    West Aussie
    Hi DonnEdwards, welcome to the forum and some very interesting banter.

    In regards to the layout.ini file within the Prefetch folder.

    I know PD uses this file in the defrag routine for optimal boot or at least what is your opinion if it's just a gimmick or actually decreases boot time?

    Do any of the other defraggers utilise the layout.ini?

    Longtime PD user here and happy with it.:)
     
  23. yeow

    yeow Registered Member

    Joined:
    Dec 11, 2006
    Posts:
    225
    Would be very interested if u smart pple find out. I was only using it out of curiousity to check with stopwatch timings. And I'll stick to stopwatch timings, much simpler.:)
     
  24. DonnEdwards

    DonnEdwards Registered Member

    Joined:
    Nov 22, 2007
    Posts:
    36
    The layout.ini file is a list of files either recently accessed or used in the startup procedure. WinXP maintains this data and uses is to optimise the file placement of these files, if you choose the right options.

    O&O Defrag has an option to use layout.ini, and PD has the option to use it as well, or maintain it, or ignore it completely and do its own thing. I am running tests on PD8 at present so I'll be able to tell you in a few days whether it makes any difference or not.

    Diskeeper claims to have helped Microsoft develop the layout.ini system (in one of their press releases or documents) and it would appear that it uses the information when it isn't using its own I-FAAST system.

    I haven't noticed a pattern yet on whether the layout.ini file helps or hinders the boot time. Either way you're looking at a maximum performance improvement of around 5 seconds with O&O and Diskeeper, which isn't really a big deal as far as I can tell. I'll see if I can find some more detailed information on how the layout.ini file works.

    See http://tweakhound.com/xp/xptweaks/supertweaks11.htm
    and http://www.raxco.com/support/windows/kb_details.cfm?kbid=574
     
    Last edited: Nov 25, 2007
  25. Franklin

    Franklin Registered Member

    Joined:
    May 12, 2005
    Posts:
    2,517
    Location:
    West Aussie
    Thanks for the reply and links DonnEdwards and looking forward to your future results.:)
     
Loading...
Thread Status:
Not open for further replies.