New User and Fast change only backups

[ThymeJTB]

Hello, I am new to IFW (as in purchased last month (April)).

I would like to know best methods when using /hash and /usemd for faster differential backups (use only /hash, only /usemd, or use both)?

What about the backup option for /del (delete the hash files used to create the backup). Does it delete the hash files used because the deleted hash files will never be needed again (e.g. for future differential backups in the chain)?

I did a test full backup with the options /hash and /usemd. Later I did a differential backup with the options /hash /del /usemd. After the Diff backup completed, I noticed it deleted all the hash files created by the full backup. Does this mean the next Diff backup will no longer have hash files to do a faster diff backup?

Last question, when using the backup strategy of Full + Differentials, would the /hash /usemd only be needed on the full backup. I ask because the differential backups are always using the full's TBI file as the base, so do you you need hash files from a previous diff backup?

[TeraByte Support]

You can use whatever method you like. Others can chime in if they like.

From the manual:

/usemd

Use this option to have Image for Windows create a hash file from the file system metadata to speed up creating a Changes Only (differential or incremental) backup. This option is only supported for NTFS partitions and relies on file system metadata to determine what gets backed up. Using this option in conjunction with Faster Changes Only Backups (/hash) can greatly decrease the time required to create the backup. This option is not supported when creating an image that is being saved to disc media (CD/DVD/BD). By default, this option is also ignored if the Backwards Compatible option is enabled.

When creating backups in Windows using this option, it’s highly recommended to use VSS
NOTE: You should only use this option if you understand its impact on backup and restore operations.

There is a whole topic on why VSS should be used here in the forums.

/del
Use this parameter when creating a differential or incremental backup to tell Image for Windows to delete the hash files used to create the backup (older unused hash files in the chain won’t be deleted).

That means whatever hash file was used to produce the new backup are deleted (typically that's done when doing incremental type backups where you don't use them again (you create a new set for the new backup)). If you only do differentials, you'd only need to create the hash on the full backup and don't delete it. You can recreate the hash, if needed, with the /hash operation.

[Bob Coleman]

ThymeJTB ,

I can't comment on /usemd except to say that based on the statement "You should only use this option if you understand its impact on backup and restore operations", I don't use it.

I do use /hash and /del. I create a full backup weekly and a chain of daily incremental backups during the week. I specify /hash when creating the full backup and both /hash and /del when creating an incremental backup. The effect is that each set of hash files is deleted after it is used.

You say that you want to create some number of differential backups based on a full backup. In that case, I would think that you would not want to specify /del when creating a differential backup so that the hash files will remain available when creating subsequent differential backups.

You ask "Does it delete the hash files used because the deleted hash files will never be needed again ...". That depends on what one does in the future. In your case, I think they will be needed, or at least used, again when another differential backup is created based on the full backup. You probably don't want to specify /del.

You also ask "I did a test full backup with the options /hash and /usemd. Later I did a differential backup with the options /hash /del /usemd. After the Diff backup completed, I noticed it deleted all the hash files created by the full backup. Does this mean the next Diff backup will no longer have hash files to do a faster diff backup"?

I would think so.

Finally, "Last question, when using the backup strategy of Full + Differentials, would the /hash /usemd only be needed on the full backup"?

I would think so.

[Brian K]

If you create a metadata differential backup, you need to retain the .@0 file. If you delete this file you can't do a metadata restore.

[ThymeJTB]

Thanks for the replies

Since a few of you and the manual say "You should only use this option if you understand its impact on backup and restore operations."
Where can I read up on what the impact is when using /usemd?

I did read the manual about this obtion, its not completely clear on what the impact is. Does metadata refer to the file allocation table, and or the archive attribute on files, etc, or something else? I read something else somewhere that made it sound like a usemd restore was like a mirror sync type restore deleting stuff not in the source and only restoring whats different.

[TeraByte Support]

viewtopic.php?p=13847#p13847

[ThymeJTB]

I read the link for /usemd option, thanks you, it was informative.

After reading the link about /usemd, it did make me realize I should point out how I do restores. I will not have any need to do any medatdata restores. When I do restores, it has always been a full restore to an empty disk. When I need to restore some files, I normally just robocopy or drag and drop the files/folders out of a mounted backup file. I am just looking for ways to make changes only backups faster.

With that said, and just doing simple restores (full to empty disks, or mounted TBI files to copy files), do you have any recommendation on speeding up changes only backups? I care more about restoring my data then restoring the OS; the OS I can always just reinstall. Its a bonus if I can restore the OS.

And also, is there a similar link that goes over the /hash option?

[OldNavyGuy]

IFW/IFL does not have the functionality to backup or restore individual files or folders.

There are other products that do.

However, you might consider either putting the OS on one drive and your data on another drive...or create a separate partition on the OS drive for your data.

Then you would be able to backup/restore the data partition only, rather than the entire system.

The downside of doing changes only backups/restores is if there is anything wrong with one of the backups [in the chain], you would not be able to do a complete restore to the time of the [latest] backup.

If you are going from an SSD to an SSD, and go "off Windows" either using TBWinRE or IFL, your backups and restores should go pretty quickly.

[ThymeJTB]

Sorry, what I mean about when I restore individual files, I use TBIview or TBIMount (depending on how many files I want to restore).

[Bob Coleman]

The full backup is not changed.

A simpler way of putting it is if any of the incremental (changes-only) backups is corrupted or missing, it results in incomplete data recovery.

Incremental is nice for backups, but restores take longer, since the original full backup, and the incrementals all have to be restored...in order.

The original poster did say the plan is to use differential, not incremental, backups.