Fragmentation vs free space

[Måç]

My XP system had been getting slow/sticky recently. The drive was almost
full (5% free) and quite fragmented. Defragging it properly was going to
take some time unless I did some clean-up to free space first. So now I have
20% free space and a defragmented drive. Works much faster/smoother now, I
am glad to say.

Fragmentation obviously occurs more when space is limited, but on a system
where one keeps (say) 20% free by regular housekeeping/archiving, frequent
defragging seems to offer little benefit.

The effects of fragmentation seem to only become really noticable as a disk
nears capacity.

Thoughts on getting the balance right?

 

[Rock]

My XP system had been getting slow/sticky recently. The drive was almost
full (5% free) and quite fragmented. Defragging it properly was going to
take some time unless I did some clean-up to free space first. So now I have
20% free space and a defragmented drive. Works much faster/smoother now, I
am glad to say.

Fragmentation obviously occurs more when space is limited, but on a system
where one keeps (say) 20% free by regular housekeeping/archiving, frequent
defragging seems to offer little benefit.

The effects of fragmentation seem to only become really noticable as a disk
nears capacity.

Thoughts on getting the balance right?

Get a larger hard drive, don't fill it so much.

 

[Måç]

Hi Rock,

I have other drives, and external storage to off load to.

I think my point was that defrag makes little *noticable* difference *until*
drive limits are approached. Kinda makes me wonder why folks bother
defragging so often, if at all.

 

[Bobby]

That has been my experience. Defragmenting a large drive with lots of room
makes little difference. On such systems, an occasional defrag (I'm talking
once every six months using the built-in defragmenter) is sufficient.

But defragementing an almost full, slow HD makes a big difference.

Bobby

 

[Rock]

Hi Rock,

I have other drives, and external storage to off load to.

I think my point was that defrag makes little *noticable* difference *until*
drive limits are approached. Kinda makes me wonder why folks bother
defragging so often, if at all.

Exactly, it's not needed as much with the current speed and size of
drives and systems.

 

[Leythos]

Exactly, it's not needed as much with the current speed and size of
drives and systems.

And that statement is flawed slightly - it's not necessary with a users
drive has a significant amount of free/unfragmented space existing on the
drive.

I have dual 160GB drives (RAID-1) in my workstation, I have about 130GB of
data on the drive. I frequently work with large files, adding/deleting
them, and do frequent backup sets on one partition. In my case, when I
edit 15K files in one day, then delete the original files and leave only
the edited file, it makes a noticeable different in performance to defrag
and consolidate free-space at least once a week.

On our file servers, we run DK9 Server version in the background, since
there is a lot of editing during the day, it keeps the servers in top
performance all the time.

So, defragmenting, is subjected to the amount of unfragmented space
already on the drive, and performance is related to unfragmented files and
consolidation of free space, and the performance gain is based on the
amount of fragmented files and fragmented free spaces in relation to how
much you hit the drive.

 

[Rock]

And that statement is flawed slightly - it's not necessary with a users
drive has a significant amount of free/unfragmented space existing on the
drive.

I have dual 160GB drives (RAID-1) in my workstation, I have about 130GB of
data on the drive. I frequently work with large files, adding/deleting
them, and do frequent backup sets on one partition. In my case, when I
edit 15K files in one day, then delete the original files and leave only
the edited file, it makes a noticeable different in performance to defrag
and consolidate free-space at least once a week.

On our file servers, we run DK9 Server version in the background, since
there is a lot of editing during the day, it keeps the servers in top
performance all the time.

So, defragmenting, is subjected to the amount of unfragmented space
already on the drive, and performance is related to unfragmented files and
consolidation of free space, and the performance gain is based on the
amount of fragmented files and fragmented free spaces in relation to how
much you hit the drive.

I knew someone would follow up with caveats...lol...figured it would
turn into the usual less filling - tastes great diatribes.

 

[cquirke (MVP Windows shell/user)]

That has been my experience. Defragmenting a large drive with lots of room
makes little difference.

This is true.

With FATxx, everything will be piled upat the "front" of the volume,
and the head travel will be minimal no matter how big it is.

With NTFS, core file system structures are stuck in the middle of the
volume, so the larger a near-empty volume, the more head travel there
will be - so it may pay to keep always-small-content NTFS volumes
small, even if there's no cluster size issues to consider.

But defragementing an almost full, slow HD makes a big difference.

Yes, because the disk usage strategy is forced to change when the file
load hits the end of the volume.

Normally, new files are created in the large mass of free space at the
end of the volume, so they are usually unfragmented unless they are
held open and updated concurrently with other files.

Once the file load hits the end of the volume, the system is forced to
create new files within the gaps between existing files, and
fragmentation becomes a bigger problem. When you delete files to free
space, these create more gaps scattered across the volume, so you
still have the same problem until a defrag pulls it all back again.

The other reason to defrag is to purge deleted directory entries.

This is particularly usefulon FATxx volumes, because FATxx looks up
directory entries in a linear fashion, from start to end. A
slowly-growing directory bloated with lots of deleted entries will
itself be fragmented and slow to access

NTFS is more efficient in that directory access is indexed via some
sort of B-tree arrangement, so that a large number of directory
entries is better tolerated than it is in FATxx.

---------- ----- ---- --- -- - - - -

Gone to bloggery: http://cquirke.blogspot.com

 

[Al Dykes]

Hi Rock,

I have other drives, and external storage to off load to.

I think my point was that defrag makes little *noticable* difference *until*
drive limits are approached. Kinda makes me wonder why folks bother
defragging so often, if at all.

Your instincts right. IMO for an ordinary user (email, MS office) on
a machine with disk that's many times lateger than needed a full
defrag after all the sw is installed and the machine is set for life.

OTOH, for someone that reinstalled large complex software packages a
defrag helps. Developers running a IDE fit this. These IDEs do heavy
IO and the user if frequently waiting for the hour glass to go away.


For apps that make huge files (ie Photoshop - 10MB to hundreds) file
can be horribly fragmented wven when there seems to be enough space.
and can slow down the user, big time. Look at the most fragmented
file report.


I do disk-disk image backups and the images can be 40GB and horribly
fragmented (thousands of segments) . This might not slow down the
backup but if the partition used for other data it will suffer.

To defrag a partition that has images in it I say you need free space
equal to twice the largest file, ie 80GB free for defrag.

I use the $$ verision of Raxco Perfectdisk. Great tech supprt for
some questions I've had.