Re: Block Corruptions on Oracle and UNIX

On Wed, 02 Nov 2005 13:41:12 -0800, g3000 wrote:

> Im seeking real world do's and dont's of running Oracle on the various
> flavors of UNIX and or Linux. What should I do to be PROACTIVE in these
> situations?
 

Give us something to bite into ... do you have a specific configuration in mind? I'm thinking about the combination of Oracle version, OS and disk. And especially whether the last is on a SAN, NAS or perhaps JBOD, and what kind of RAID, if any.

That said, some of the proactive DOs are fairly obvious:

• keep the Oracle and the OS patched to reasonably current patch level;
• keep on supported versions, (and that excludes the extended support);
• use RMAN and it's built-in block oriented capabilities;
• I favour running a regular dbv, but RMAN somewhat obviates that;
• do use a non-journalling filesystem;
• look at RAW (or RAW+ASM in 10g);
• look at reference manual for more info around:
  • You should set DB_BLOCK_CHECKING to FULL if the performance overhead [of between 1-10%] is acceptable;
  • set DB_BLOCK_CHECKSUM to TYPICAL.

The last two are system parameters with Oracle recommendations.

-- 
Hans Forbrich                           
Canada-wide Oracle training and consulting
mailto: Fuzzy.GreyBeard_at_gmail.com   
*** Top posting guarantees I will not respond further ***