Re: *SAVFs -vs- Direct to Tape performance.

[Larry Bolhuis <lbolhuis@xxxxxxxxxx>]

Rob,

 Most recently I did this on an 825 with 4 Processors, 6G (I think) 
memory and 50 15K rpm drives. Tape drives on the box are 3581s. Even 
paralleling the 3581's used about 10% of the CPUs. With four parallel 
*SAVFs we push the CPUS to 60+% and that's good for now.  My CL is set 
to support up to 16 parallel saves.

  In one case we even break the largest library into four parallel save 
pieces.  The program that copies the *SAVFs to disk also authors a CLP 
that will restore everything and saves it at the beginning of the tape.

 - Larry

rob@xxxxxxxxx wrote:

> Larry,
>
> There was quite a discussion on whether or not save files were any faster. 
> And lots of people reported the opposite.  The only new argument you 
> bring the table is the parallel concept.  That might be significant.
>
> I'd like to know the system specs:  Model and tape drive, that you 
> enhanced.  Too often I hear stuff that someone discovered in V1R2 on a B10 
> and take as gospel today.
>
> Rob Berendt
>  

--
Larry Bolhuis                   IBM eServer Certified Systems Expert:
Vice President                    iSeries Technical Solutions V5R2
Arbor Solutions, Inc.             iSeries LPAR Technical Solutions V5R2
1345 Monroe NW Suite 259          iSeries Linux Technical Solutions V5R2
Grand Rapids, MI 49505            iSeries Windows Integration Technical 
Solutions V5R2
                               IBM eServer Certified Systems Specialist
(616) 451-2500                    iSeries System Administrator for 
OS/400 V5R2
(616) 451-2571 - Fax              AS/400 RPG IV Developer
(616) 260-4746 - Cell             iSeries System Command Operations V5R2