Defrag and incremental backups

Another experiment
Addendum to my post: Apropos data consolidation

Same test as described above, with the exception that the partition has been
defragmented with O&ODefrag prior making the full backup:

1) Defrag with O&O Defrag ("space" mode used)
2) Full image of the defragmented partition: 1.36 GB
3) Restore this image
4) Incremental image of the same partition: 117 MB

Interessting! This is a completely different result. Looks like it makes a difference, which defragmentation tool is used. O&O Defrag consolidates data and apparently fits well with TI: It seems, that in this case TI restores most data into the original clusters. There is much less discrepancy between the sector map in the tib file and the effective sector allocation on the restored partition. This keeps the incremental small.
At least, this is my explanation; perhaps somebody knows some technical details? Would be nice, to hear the opinion of the TI support team about 'defrag and incremental images'

BTW, this is not an ad for O&O Defrag. It's just the tool I have

 

Not really technical, but there are different definitions of fragmentation. There is file fragmentation and disk fragmentation.
If each file on your system is contained in a contiguous piece, there are programs that will report your drive as “not fragmented” because the files are not fragmented. Telling it to defrag the drive will not result in much change.
If, however, these files are scattered all over the disk, your disk could be said to be fragmented. Other progs will report this disk fragmentation and defrag the disk space as well as the files. I think older MS defraggers offered the choice of file or disk defragging. It sounds like O&O defrags both the files and disk space.
If the files are not fragmented, but the disk is, creating an image and restoring could result in a vastly different set of “in use” sectors. But if the disk, and not just the files, has been defragged, an image/restore might look very much the same as the original. This matches what you’ve seen.

 

Thanks for this information. Before I started the first experiment (post: April 20th, 2006 09:03 AM), files were defragmented (with WinXP defrag tool). The disk didn't look much fragmented. This is at least, what I could see in the block diagramm of the defrag tool. But it is feasible that there were gaps with unallocated sectors, which were not displayed in the diagramm. Maybe one little space will cause already a huge dislocation of following data sectors during image restoring.

 

Thanks JothiS for your tests and information. I'm actually using O&O ("space" mode) too and notice real performance improvements in some areas. My system mostly holds many small files (I develop in Java, PHP etc) but also keeps some larger (high res images) and really large files (videos, CD/DVD images, game files).

The performance increase is especially noticable e.g. when starting the Java app Eclipse, opening/saving large images in Gimp/Photoshop, copying large files around from one disk/partition to the other and on startup of the XP System.

In every days work there is no noticable difference, e.g. loading/saving a file from within OpenOffice/Eclipse is just fast.

So, I'm not such a defrag enthusiast, but still do it from time to time: once a week. Especially because of the "disk defragmentation" (hi TheWeaz, good explanation), O&O is good at consolidation of free space on the disk whereas the windows defrag "tool" seems to ignore that, or is just unable to consolidate free space effectively. Consolidation of free space reduces further defragmentation when adding new files.

And after defrag, I always do a full backup with TI. In the meantime I do differential TI backups, to reduce potential restore activity to max 2 files.

My full backup from the internal disk (3 partitions) to an external USB drive with plenty of space (400 GB for ~ 200 $/€) uses 30 GB backup space (44 GB data to backup on the internal 100 GB disk) and needs approximately 25 minutes.

The differentials start out at 0.2 GB to 1 GB (first day) and increase up to 2.0 to 4.5 GB (last day before next full backup). Notice, that they ALL take almost exactly the same backup time as the full backup: 25 minutes.

When working (Eclipse/OpenOffice/Gimp) the running backup is almost not noticable. TI uses around 25 % processor time, my typing/mouse moves take about 0.0000001 % processor time. Loading/Saving files, checking files out from CVS via internet etc. don't "feel" to be any slower.

So the O&O defragmentation combined with the TI backup is a perfect solution for me. I just ignore both most of the time (having scheduled their activity), being sure, to have a backup all the time and a speedy disk most of the time.

Ok, when playing a modern shooter game at backup time, that doesn't match very well. TI's activity reduces my frame rate drastically, and the backup takes forever. But what? How stupid can one be?

regards, Georg

 

Dear all,

I just got an email from the Acronis Support Team. Acronis has solved the defrag problem in the newest build 3567. That's good news ! (I guess, you English speaking guys have this build already a loooong time!)

Acronis has implemented much more new features in build 3567. Have a look on the excerpt from their email. It sounds very good. Hope, the BartPE plugin will work. (Sorry, it's in German - but you'll surely find the infos on the Acronis site, too)

Have a nice day!

 

I used a couple of language translation websites and by consolidating them the best translation I could come up with is:

The update has some new functions, which are not at first sight evident, but the jump in the Build number and the size of the installation file to explain.
- Acceleration of incremental and/or differential Backup (with it the Defragmentierung does not affect itself any longer negatively)
- User-definable standard options for Backups
- User-definable standard options for restorations
- Treatment of gemounteten image archives
- Boat-able backup-media
- Boat-able ISOs and RIS packages
- Option, which Acronis components the emergency medium should contain
- BartPE Plugin
- Support for Windows XP x64 edition
- Support of GPT non removable disks (non removable disks with GUID partition Table)

I haven't changed any translated words hence words like Boat-able.
I guess you can interpret such words yourself.

 

I have been pleased too early.
I can't verify the statement of the Acronis Support Team: I have repeated my tests with build 3567 and the results are still the same; no improvement at all! Defragmentation as well as storing/restoring of a full backup lead to very big incremental images.