Re: Q: Is RAID/striping good with Oracle???

From: KELLEY SCOTT A <skelley_at_hsc.hac.com>
Date: 1996/08/22
Message-ID: <4vijs0$r79_at_hacgate2.hac.com>#1/1


YES! A Storage Array is the method of choice for a Sun. Not only does it support RAID 0+1 for your database (allowing you to stripe across every disk in the box), the NVRAM write cache will increase your write throughput dramatically. It also supports hot spares. I cannot stress enough the benefits realised from an array, it is far better than just-a-bunch-of-disks, even if you are using Online Disk Suite. Buy yours today.

In response to the earlier post. RAID 0+1 (mirrors and stripes)i s what you need to get the best performance. RAID 0 protects your data from a disk failure and give additional platters to read data from. RAID 1 spilts the logical volume among multiple physical devices to reduce i/o bottlenecks. Stay the hell away from RAID 5 for DB applications.

p.s. I'm sure third party vendors have comparable devices to the Sparc Storage Array for Sun's and other platforms.

            Scott Kelley, Oracle Corp. (skelley_at_us.oracle.com) "Oh! My eye!"

In article <321A1D99.7AE1_at_worldnet.att.net>, netac <netac_at_worldnet.att.net> writes:
>Gary Assa wrote:
>>
>> Would RAID and striping be a good thing with Oracle 7?
>>
>> Being that tables should be on different drives from indexes, etc.,
>> is RAID defeating the purpose? Or is it an insignificant amount
>> balanced by the speed of RAID?
>>
>> I am really not up on the way RAID works.
>> --
>> =========================================================
>> http://www.li.net/~gsa/index.html
>> This is my signature file, not part of this mail message.
>
>RAID does not necessarily preclude keeping tables and indexes on
>separate disks (or sets of disks). I too am interested in any comments
>from people on this topic. We are thinking of upgrading our SPARC 1000
>to include a SPARC storage array with more drives for RAID, and I would
>be interested in hearing about people's experiences with this
>configuration....
Received on Thu Aug 22 1996 - 00:00:00 CEST

Original text of this message