Re: How would you layout the files?

From: Bill Ferguson <wbfergus_at_gmail.com>
Date: Mon, 8 Dec 2008 17:07:03 -0700
Message-ID: <4025610e0812081607t3c6451b6q2e3f1283690cdfbb@mail.gmail.com>


I'm running on a very similar setup. Dell Server, Windows 2003 Server, etc.

I have the drives setup for RAID 5 with two partitions, Drive C and Drive D. I have 50GB set aside for the C Drive (primarily the OS), but I also multiplex the control files on the C Drive.

My database is similar in size, around 100GB, and everything is on the D Drive (1.3TB), Oracle, data, etc. There is a barely noticable performance hit with this configuration, but it's cheap (we replace the server every 4 years for whatever we can get get for $10K).

I've had a drive die twice on me, and both times I was able to replace to drive and rebuild it without problems. The system continued to run until the new drive arrived, then I took the system offline while I performed the rebuild, about 4-5 hours.

I run a nightly full database RMAN job and store the info in the control files. I also have about a 4TB tape library which backs up everything each night.

I only have a small number of users though, so this arrangement works well for us, especially with drastic budget cutbacks. We requested $125K in non-salary budget expenses this year and received $20K, so they get what they pay for.

If the clients machine and database only has 20-50 users and they can accept being down for a day, maybe two at the most, the configuartion should work, though when they can afford it, I would suggest they get larger drives, or at least look at getting the Dell PowerVault for additional hard drives. Storage is the one area my original configuration of 4 years ago kept biting me in the backside, I only had about 500 GB at that time, and some update statements would fill the flashback recovery area with redo logs and hang Oracle, so I would have to manually delete the files, turn off flashback, run an impromptu RMAN session to clean the obsolete's out, then turn flashback on again, etc to get the database to complete the updates.

--

Received on Mon Dec 08 2008 - 18:07:03 CST

Original text of this message