Re: Oracle Standard Edition & RAC

From: Job Miller <jobmiller_at_yahoo.com>
Date: Thu, 4 Jan 2007 21:08:53 -0800 (PST)
Message-ID: <20070105050853.31711.qmail@web53913.mail.yahoo.com>

>I'd be interested to hear about this as well.

IBM Joint Solutions Center did some decent testing of a two site, two storage system, 2 node cluster and put together the detailed tech architecture and ran through the following tests which essentially boiled down to crashing each individual component of the overall system (controller, storage arrays, network interfaces, servers, instances, etc) to see how it responded. Since it was only 2 nodes, 2 sites, they didn't do more elaborate testing of killing parts of the surviving cluster while it was reconfiguring (as kevin suggested in an earlier thread)

  Power off Server
  Unplug client network
  Unplug Interconnect cable
  Halt –q on Linux
  Kill oracle Instance
  Kill ASM Instance
  Unplug fiber attachment
  Power off one FastT
  Fiber card Failure

  Disk failure - LUN_ocr 
  Disk failure - LUN_voting
  Disk failure - LUN_data

on 10gR1 and 10gR2 and documented the results. it is all in a ppt I can send your way if you are interested.. I am sure you have also seen the stuff on OTN about it in the MAA section of OTN.

10gR1 had some issues/limitations, in 10gR2, the full Oracle stack worked when you put a third voting disk in an external location.

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Jan 04 2007 - 23:08:53 CST