cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted

Fragment Size

Dear Experts,

I tried to grasp the concept of Fragment size in policy but didnot get the undertanding properly Smiley Sad

Could somebody pls explain?

Thanks so much,

 

1 Solution

Accepted Solutions
Highlighted
Accepted Solution!

I tried to grasp the concept

I tried to grasp the concept of Fragment size in policy but didnot get the undertanding properly Smiley Sad

Fragment size is actually a Storage Unit setting - not policy setting.

Reduced (smaller) Fragment size has the advantage that restores are normally faster.

It works like this:
On tape, the default fragment size is 1TB. 
So, if you have very large filesystem backup with lots of small files, the backup will be written in a single tar-file. If the backup is bigger than 1TB and media has space left, a file-mark will be written (Fragment1) after 1TB and a new Fragment will be started.

If you now want to restore a single file, NBU knows that FileX was written in Fragment 1. When the tape is mounted, the entire 1TB fragment must be scanned for that 1 file.

If the Fragment in STU config was reduced to something like 5000MB, it means that NBU will write a filemark (fragment) after each 5GB of data written.

If you now want to restore FileX, NBU will know which fragment number this file belongs to (eg Frag 3).
The bptm process will now skip fragments 1 and 2 and position tape at Frag 3. Only this 5GB fragment must be scanned until FileX is found.

I have seen how restores of single files was cut down from a couple of hours to a couple of minutes with reduced fragment sizes.

The backup will take slightly longer, because data transfer must be paused in order to write the filemark. This should not take longer that a couple of seconds and in the long run not really make a difference in backup time.

The advantage of reducing fragment size of disk used for GRT backups is that only the required fragment will be NFS mounted to search for individual items.

Hope this helps.
 

View solution in original post

5 Replies
Highlighted

The Reducefragmentsize

The Reducefragmentsize storage unit setting specifies the largest fragment size that NetBackup can create to store backups.  (From Admin Guide)
 
for example you have backup size of 100 GB and you set the fragment size as 50 GB in storage unit, it will write the backup images as 2 files ( 2 fragments) in the storage unit.
 
in below post Nicolai reply has the recommended settings for this.
 
see below
http://www.symantec.com/business/support/index?page=content&id=HOWTO56051
 
https://www-secure.symantec.com/connect/forums/how-determine-best-netbackup-fragment-size
Highlighted
Accepted Solution!

I tried to grasp the concept

I tried to grasp the concept of Fragment size in policy but didnot get the undertanding properly Smiley Sad

Fragment size is actually a Storage Unit setting - not policy setting.

Reduced (smaller) Fragment size has the advantage that restores are normally faster.

It works like this:
On tape, the default fragment size is 1TB. 
So, if you have very large filesystem backup with lots of small files, the backup will be written in a single tar-file. If the backup is bigger than 1TB and media has space left, a file-mark will be written (Fragment1) after 1TB and a new Fragment will be started.

If you now want to restore a single file, NBU knows that FileX was written in Fragment 1. When the tape is mounted, the entire 1TB fragment must be scanned for that 1 file.

If the Fragment in STU config was reduced to something like 5000MB, it means that NBU will write a filemark (fragment) after each 5GB of data written.

If you now want to restore FileX, NBU will know which fragment number this file belongs to (eg Frag 3).
The bptm process will now skip fragments 1 and 2 and position tape at Frag 3. Only this 5GB fragment must be scanned until FileX is found.

I have seen how restores of single files was cut down from a couple of hours to a couple of minutes with reduced fragment sizes.

The backup will take slightly longer, because data transfer must be paused in order to write the filemark. This should not take longer that a couple of seconds and in the long run not really make a difference in backup time.

The advantage of reducing fragment size of disk used for GRT backups is that only the required fragment will be NFS mounted to search for individual items.

Hope this helps.
 

View solution in original post

Highlighted

Hi Marianne, Thats what i was

Hi Marianne,

Thats what i was looking for...It cleared all my doubts.... Thanks so much.. U are also a best teacher Smiley Happy

 

Highlighted

Thanks so much Nagalla......

Thanks so much Nagalla......

Highlighted
Moved:

Re: I tried to grasp the concept