Re: Oracle db & Fault tolerance Implementation

Yes. RAID is excellent for read and terrible for write. We put our redo logs on internal SCSI disks because all you do is write,write,write to these logs. With archiving on we haven't lost a bit of data so putting them on SCSI wasn't too risky. The only performance tuning we've been able to do with RAID is understand how many disk controllers are controlling what disks and try to spread the load between them for your "hot" data. With RAID, practically all our tuning tricks with disks are out the window! ==========Wong Yeow Kiat, 12/12/95========== Hi there,

I'm going to implement the Oracle database in the Alpha server with Raid '5', fault tolerance feature (The
server will be installed with multiple "physical" hard disks). Information that I have gathered is that the Oracle database requires different hard disk partiton
(different disk controller could make it better) for storing the data,
rollback segment and redo log file
separately to avoid I/O contention and for better system performance. With the Raid 5, "distributed data guarding" implemention, there will be only one logical partition
(Correct me if I'm wrong).

Is there any implication on the above setup?

Appreciate for the help.

regards,
Yeow Kiat (S'pore Petroleum Co./Computer Resources Dept.) Received on Fri Dec 15 1995 - 00:00:00 CET