UDefrag Method and Dir Close to MFT

Discussion in 'backup, imaging & disk mgmt' started by TraumaDoc, Jan 30, 2011.

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

    TraumaDoc Registered Member

    Joined:
    May 24, 2009
    Posts:
    75
    I've gotten annoyed with defraggers and VSS and boot-time screwing with restore points and so on, so I've given up on restore points, etc.

    Anyhow, in a previous thread I noted how after I moved the boot-time files to the start of the drive, that the directories were still being stuck in the middle of the drive when I did a Windows-mode defrag and placing directories close to the MFT. The UD developer wrote a comment or two about it but I think I just realized a piece of information he neglected to add that would have made it more clear to me.

    So, is the following assumption correct? Is that when I tell the directories to be placed near the MFT that the directories are placed BEFORE the MFT and not after it or the NTFS boot-time tab in settings files? I can now understand how the directories could not be placed before the MFT and get stuck somewhere else if the MFT (and NTFS boot-time files) are placed at the first available cluster (# 32). Is this correct?

    So, then, if it is correct, what would be the optimal placement for the beginning of the MFT (and boot-time files block noted in the boot-time defrag settings tab) so as to allow more than adequate supply of space for the directories of an entire system drive (of course taking into account a large number of directories and directory growth) ??

    Would someone mind giving me a clue on this?

    I've also gone back to using only 4 KB cluster sizes and not the 16 KB needed for "proper" VSS use. Udefrag once again screwed me when I did a boot-time defrag with 16 KB cluster sizes and fried my MFTMirr and "forced" it to be re-sized to 4 KB and not 16 KB.

    So, with this information, where should I place the MFT and boot-time block of files so to allow more than just enough space for directories to be placed before/near/close to the MFT ? I'd appreciate a specific notation, not just someone telling me to calculate cluster sizes and directory size, etc. For example, my 365 GB system partition currently has more than 15,000+ folders and yet I'm using only 21 GB of space on the 365 GB partition.

    What would be a good way for this ?

    This assumes of course, two things:

    1 - that yes, directories are placed before the MFT in this setup
    2 - that placing the MFT and boot-time files near the start of the partition are best

    Help, please.

    Thanks all !!

    :p
     
  2. crofttk

    crofttk Registered Member

    Joined:
    May 15, 2004
    Posts:
    1,976
    Location:
    Eastern PA, USA
    No elegance in the way I do it. After I've done on online defrag I just look at the color coded blocks on the drive map ( I switch directories to pink so I can them easier) and spot where I want MFT to start, allowing for the number of blocks that I see my directories will occupy. Then I left click on the target block to activate cluster viewer and read the beginning cluster number at the top of the cluster viewer display. I then specify that number in the starting cluster field on the boot time defrag tab.
     
  3. TraumaDoc

    TraumaDoc Registered Member

    Joined:
    May 24, 2009
    Posts:
    75
    Sorry, but that's just not simplifying it for me.

    How can I know how many "clusters" are required for the directories ? 1 cluster (4 kb) = 1 Dir ? Currently, my directories are stuck in the middle of the drive - how can I be sure how many blocks/clusters I need free at the start of the drive, starting at cluster 32 before I can allow the MFT and boot-time files to be placed ?
     
  4. crofttk

    crofttk Registered Member

    Joined:
    May 15, 2004
    Posts:
    1,976
    Location:
    Eastern PA, USA
    Simply look at the drive map and count how many blocks on the map your directories are occupying - I'm assuming at this point your directories are grouped together and you can see by color code where they are. The color codes are more useful in distinguishing among directory, MFT, mft-reserved, boot-time and metadata if you view them immediately after an online defrag. Here's how mine looks at the moment:
    UD3a.jpg
    My directories are grouped after the high performance (boot-time) files and a small group of most recently used files and are followed by the mft, metadata, page file, and hibernation file. (And, yes, I have only a very small MFT-reserved area, the yellow.)

    Assuming you have used the cluster viewer, arrange the drive how you like and click on the block where you want MFT to start, after advancing along the drive however many blocks you want to allow for your directories to go in front of MFT, and read off the cluster number in the cluster viewer.

    If you feel the need or desire to be more precise and controlling than that, there's no escaping doing the math, IMO. Doing the math, however, has it's own aspect of simplicity that you may or may not appreciate.

    Back to your OP - as far as how much extra space to allow for additional directories in your chosen spot, I think you're better off just experimenting with it. I would guess that additional space would just get filled in with other data, especially near the front of the drive ,but you can displace that data each time you defrag online, without having moved the MFT, and have your directories grouped together in roughly the same spot.

    Hope that at least gives you food for thought if it doesn't further simplify it for you.
     
  5. TraumaDoc

    TraumaDoc Registered Member

    Joined:
    May 24, 2009
    Posts:
    75
    What I chose to do was check the total number of directories on the drive at the current time, then doubled it. So, my C main drive has 15,405 directories.

    I then set the MFT to start at 30,000

    Another partition has only 300 directories, so I just set it to 1000. In the view of UD, one block was equal to 1700 clusters or so, so that drive has only movies, so felt 1000 was adequate.

    Under this scenario, the directories are clearly visible at the start of the drive, before the MFT as I wanted.

    I assume this is a fair and safe method achieving the necessary goal.

    In my opinion, there should be some sort of warning in UD to say that there is no space available to move the directories close to the MFT when you choose that option. Disagree ?
     
  6. crofttk

    crofttk Registered Member

    Joined:
    May 15, 2004
    Posts:
    1,976
    Location:
    Eastern PA, USA
    Not at all - seems it should be simple enough for them to check that ahead of time.
     
  7. TraumaDoc

    TraumaDoc Registered Member

    Joined:
    May 24, 2009
    Posts:
    75
    You're joking, right?

    You REALLY think most users know what a cluster, block, etc. are? Many people can't even tell you the true definition of a bit, byte, kilobyte, etc.

    A feature that cannot be completed as instructed in an option should be warned against.

    That's one of the most asinine comments I've seen.
     
  8. crofttk

    crofttk Registered Member

    Joined:
    May 15, 2004
    Posts:
    1,976
    Location:
    Eastern PA, USA
    Dear TraumaDoc:
    Please forgive me for not being sufficiently explicit in my oversimplified answer to you. Let me rephrase: "NO, I DO NOT DISAGREE WITH YOU AT ALL, IT SHOULD BE A SIMPLE MATTER FOR UD3 PROGRAMMERS TO CHECK AHEAD OF TIME IF THERE IS A CONFLICT AND TO THROW A FLAG TO THE USER."

    Asininity abounds where one is inflexible in their semantic and syntactic interpretive skills.

    P.S. RELEASE THE KRAKEN!
     
    Last edited: Jan 31, 2011
  9. DOSawaits

    DOSawaits Registered Member

    Joined:
    Dec 11, 2008
    Posts:
    415
    Location:
    Belgium
    I'm not a specialist on this matter, but logical thinking drives me to the idea that the "ideal spot" for MFT and directories should be about 1/3 (one third) into the currently occupied zone. Placing MFT and/or directories as close as possible to the start of the disk will result in way too much head movement, and will actually slow things down, especially when a disk is quite filled already. In my view most current free defraggers do a nearly perfect job OOTB. All worying about how to squeeze just that little more is a waste of time imho.
     
Thread Status:
Not open for further replies.