Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.misc -> Re: How critical is RAC experience

Re: How critical is RAC experience

From: Michael Squires <mikes_at_cs.indiana.edu>
Date: Fri, 29 Apr 2005 15:36:15 +0000 (UTC)
Message-ID: <d4tk9f$596$1@rainier.uits.indiana.edu>


>hpuxrac wrote:
>
>> RAC is just oracle on multiple different nodes simultaneously. You
>> just administer 1 database at a time, like always.

Speaking as a user of an RAC cluster (10g) there was a significant learning curve. One major problem is that the consultant who set up the original cluster was not as knowledgeable as we thought, and there was one major crash with major database corruption as a result.

There are two major sets of users; one set is basically transactional, the other (me) creates and maintains databases with 10,000 to 500,000 records, using Access and Toad to view the data and Discoverer (DS version) for analysis. Access wouldn't work until we found the 10...3 Instant Client (the 9i client worked, but caused the aforementioned crash and database corruption). Discoverer was very slow until the local admin was able to get into the RAC and tune it; to properly tune it they're going to have to reinstall it this summer (we're supposed to be able to connect to "our" network node, but right now we get assinged to a node by the RAC).

There is also an existing bug which causes updates with more than 5,000 records to fail when the database merges the results of an SQL*Plus session with the master copy. This is not affected by the number of records which are committed by the SQL code, makes no difference if you commit one record at a time or all of them (in your code).

After all the tuning work, and some network upgrades, performance is good - better than the dual CPU Xeon server which ran 9i for us.

On the other hand, my home database server running postgreSQL 7.4 (under FreeBSD 4.11-STABLE on older server hardware, Supermicro P6DGH) has a better uptime record than the office's RAC cluster - I'm at 100% for the past year, and the RAC cluster has been down for several days since it came up about six months ago.

Mike Squires
RAC user

-- 

Mike Squires (mikes at cs.indiana.edu)     317 233 9456 (w) 812 333 6564 (h)
mikes at siralan.org             546 N Park Ridge Rd., Bloomington, IN 47408
Received on Fri Apr 29 2005 - 10:36:15 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US