tag:blogger.com,1999:blog-5124641802818980374.post7997266067188834895..comments2023-09-06T11:58:30.752+02:00Comments on Oracle related stuff: Big Nodes, Concurrent Parallel Execution And High System/Kernel TimeUnknownnoreply@blogger.comBlogger1125tag:blogger.com,1999:blog-5124641802818980374.post-85794501762223409992016-02-25T10:07:10.599+01:002016-02-25T10:07:10.599+01:00Dear Randolf,
thanks for sharing the resolution o...Dear Randolf,<br /><br />thanks for sharing the resolution of the problem observed. I'm glad to see, that a relatively simple configuration change improves the situation a lot.<br /><br />My 2 cents in such situations: anyone running into such situation (PX seems to deliver very little). Please check MOS for known problems. Without knowing the root cause of a problem, it is not possible to judge, if the applied change lets you land on an island of stability.<br /><br />Generally speaking: parallel execution in multi-user environments require great care. Make sure, everyone gets the resources without blocking the others. I've seen too often an optimistic approach ie parallel degrees far too high or PX contention.<br />So watch out for any events in the "contention" class and/or being "PX" related.<br /><br />It would be great to see, how your use case works in 12c. Have you had a chance to test that too?<br /><br />Kind regards<br /> Thomas<br /><br />p.s. I'm an Oracle employee. These are my personal thoughts.Thomas Teskehttp://www.oracle.comnoreply@blogger.com