Re: Metrics for Swingbench benchmarks
Date: Fri, 14 Mar 2008 11:25:16 -0400
On Fri, Mar 14, 2008 at 10:44 AM, John Hallas <john.hallas_at_bjss.co.uk> wrote:
> The main misunderstanding I have is the difference between 95% and 90%.
> The 2nd node is available but not in use. I know there will be heartbeat
> and other maintenance work going on and I am also aware that the cache
> fusion will keep the 2 SGAs in sync, however I have been searching for more
> information on what is happening in the preffered/available modes and any
> thoughts appreciated.
When you say "preferred/available" are you referring to the "Service" you created? If so then the only effect of this is that all connections are going to the preferred node - however the other node is up and running and fully part of the cluster.
One thing that comes to mind right away - even though your second node is "not in use" it still maintains a complete mirror of the GRD (global resource directory - things like cluster-wide locks). Also, there's a good amount of additional messaging related to SCN generation. Just some shots in the dark, but I wonder if you could statically force all lock masters to the preferred node and decrease SCN traffic by tweaking something like the MAX_COMMIT_PROPOGATION_DELAY init param? That might get you some of those percents back...
PS - I'm loving this whole thread because it's a perfect example of the bogus hype around RAC. Clustering is powerful technology and honestly, a lot of applications will scale well with no modifications on RAC. (The well-designed ones.) But clustering is complicated and if a sales person gives the impression that you can just run any application on RAC to make it magically faster -- baloney!!!!
-- Jeremy Schneider Chicago, IL http://www.ardentperf.com/category/technical -- http://www.freelists.org/webpage/oracle-lReceived on Fri Mar 14 2008 - 10:25:16 CDT