You did not mention how much memory you have in your 939KS.  You may be
able to "earn" back (plus more) and loss in speed that an AutoRAID 12H
introduces by adding additional memory to the HP e3000.

The data locality for your application makes a significant difference on
the speeds you will be seeing from the AutoRAID box.  Data locality is
dependent on the size of the database, the organization of the database
(linkages), how often you clean up and optimize (repack) the database.

Halberggsb wrote:
>
> We plan to buy an AutoRAID 12H to use it with our productive ERP (that means
> not our test systems). The vendor told us that the I/O performance would be as
> least as good as now with our simple disks, but in the meanwhile I've read
> other things here in the newgroup. Especially I am alarmed for a special
> I/O-intensive batch which fits for the night; it would be bad if he needed 10 %
> more time.
>
> We have seen that the 12H could run with RAID-1, because our productive data is
> less than the half of it's raw capacity.
>
> --> Therefore, above all, I am interested in the difference of the
> 12H-performance compared with the performance of simple disks.
>
> A HASS with full mirrored disks, two SCSI-controlers and MirrorDisk/iX would be
> more expensive than the 12H that we could buy.
>
> As we need more disk capacity now, we would buy some simple disks now and would
> transit to a mirrored system later, if the performance risk were too high.
>
> The details: Now in our 939KS with 70 users we have one F/W-SCSI-Controler with
> 3 9GB disks, 1 4GB disk and 5 2GB disks, totally 41GB. The AutoRAID has 8 9GB
> disks; we want retain the 4 biggest old disks for the test systems and similar;
> the productive MPE accounts for the 12H need 20-25GB.
>
> Who could image how performance will change, or could give advise?
>
> Thank you very much.
>
> Gerhard Gross
> HALBERG GUSS GmbH
> Kirchstr. 16
> D-66130 Saarbrücken
> [log in to unmask]