Hi,
We're working on optimizing a large Oracle database on a couple of
GS140's running Tru64 Unix 4.0F (with all recent patches applied). We're
using Qlogic ISP 1020A SCSI controllers (4 per system) connected to 4
shared MTI 9000 arrays (utilizing TruCluster and OPS for
failover/balancing), using raw partitions for all database storage. In
addition we have all filesystem-based storage on local disks hanging
off of 53C810 based controllers (note: these are all Compaq parts, I'm
just really bad with Compaq part #'s :) ). I'm trying to figure out the
optimal block size to use for striping across the MTI arrays, and in the
process trying to find the maximum block read/write sizes that Tru64 will
support at the hardware level with the 1020A's in this configuration. Also
any other configuration tips from folks that have dealt with the same or
similar setups would be appreciated.
RTFM is an answer I'll gladly accept, given someone can tell me where to
find the FM :) At this point I'm working from the Cary V. Millsap paper
("Configuring Oracle Server for VLDB") and playing with configuration
options based on the recommendations therein.
Thanks in advance,
-Tymm
Received on Fri Mar 31 2000 - 18:58:13 NZST