demultiplexing is extrememely expensive in terms of time - each client has
to be read and written in sequence from scattered chunks.  The time increase
is more than just the mux factor (4 in your case) because the mechanical
delays added by the seek-to-fragment are an issue.  

Why are you thinking about doing it?  I moved a duplication job from
literally days to hours by not demultiplexing. (Actually, I started
duplicating specially created non-mux'd backup - demux-ing at the source
rather than the destination.)

Since you're duplicating 'all' your backups - have you considered writing
two copies at backup time using the in-line tape copy that's part of vault?

-m

-----Original Message-----
From: Mark Kilpatrick [mailto:mark.kilpatrick AT Redstor DOT com]
Sent: Friday, February 20, 2004 5:41 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Netbackup Vault - Duplication


Hi,  do you have any approximate stats for duplicating backups using Vault.
I intend to duplicate all of a previous nights backups which are on one
tape. My multiplexing level at the time of backup is 4. I have LTOgen2
drives.

What kind of time difference should I see in the time taken to backup
compared with the time taken to Vault. The vault job is de-multiplexing.
For example, if a backup of four clients to one tape with multiplexing of 4
takes 1 hour - how long should the vault duplication take?

Thank you,

Mark Kilpatrick


_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu