----------
> From: Daniel Thompson <thompsod AT usaa DOT com>
> From: Daniel Thompson <thompsod AT usaa DOT com>
> To: ADSM Answers <adstarsw AT vnet.ibm DOT com>
> Cc: Ted Kicey <kiceyt AT usaa DOT com>
> Subject: Parallel Incrementals on same node.
> Date: Tuesday, September 23, 1997 8:54 AM
>
> Greetings ADSMers,
>
> Last week there was a discussion thread on how to speed up incremental
> backups.  In addition to the discussion that went on I stated I would do
> some testing on simply running multiple backups on the same node at the
> same time.
>
> Somebody pointed out that ADSM builds the "blueprint" for the files to be
> backed up at the beginning of the backup therefore concurrent backups
would
> each process all of the files needing backing up.  Based upon this, I was
> curious if the slowincremental option would bypass this pre-processing
and
> allow concurrent backups.
>
> Well, the answer is that concurrent backups using slowincremental WILL
> speed you up, but only a little and it is chock full of inefficiency.
> Specifically, the incrementals will not re-backup a file once it has
> successfully transferred to ADSM.  However, if the file is in the process
> of transferring, multiple backups will process it.  Therefore, almost all
> of the parallel processing is wasted and only occasionally will one
process
> start on a file and complete it before the other processes can get to it
> and begin transmitting.
>
> I am not aware of any other methods that would facilitate simply running
> parallel incremental backups on the same node.
>
> Thanks,
> Daniel Thompson
> Storage/Automation Management - USAA
> thompsod AT usaa DOT com