FREE STUFF !!! FREE STUFF !!! FREE STUFF !!!
At the last week’s Expert Oracle Exadata virtual conference both me and Kerry touched the topic of how smart scan (and direct path read) decisions are done for each scanned segment during the SQL execution runtime – and not by the optimizer during the optimization phase.
I got a few follow-up questions about this yesterday and I also recall a similar question after my Understanding Exadata Performance Metrics presentation I did for UKOUG in London in April.
I think this is a topic which deserves some deeper coverage and so I decided I’ll do another secret hacking session on Tuesday 9th August (2011) 9-11am PDT.
The topics (hopefully) covered will be:
- How do full table scans work
- How and why do the “ORA-8103: object no longer exists” happen
- How does Oracle (11g) decide between a buffered full table scan and a direct path read scan (plus smart scan in Exadata)
- What’s the difference between an object_id and data_object_id?
- and more!
Note that while I do touch the Exadata topic a bit too, this session is generic and useful for anyone full scanning tables on any Oracle database… :)
This event is free and you can sign up here:
P.S. I will record the event and publish the video too. So if you’re in Australia or Hong Kong, no need to stay up late ;-)
Hello,
I may have missed something, but what happened to that “most complex ever performance issue” you mentioned a while back?
Cheers,
Simon
@Simon Palmer
Simon, I delivered this at Hotsos Symposium this year, but haven’t done it online yet. I’m thinking for doing that as a virtual seminar (I have many more very complex issues to talk about :-)
Aaah yeah, ah well. Hope you do it online sometime. Cheers, Simon
Tanel,
thank you very much that you do not forget about folks in Australia!
Look forward to watching the session!
Hello:
–> Tuesday 8th August (2011)
a typo?
Thanks for the sessions…
@AS
Bummer – yep typo, Tuesday 9th Aug!
Thanks – great session!! :) Especially enjoyed seeing the bootstrap process – hadn’t learned about that before.
One follow-up question about direct path vs scattered reads. Other than reading into a different memory location are there any differences in how the read itself happens? Any latch or SGA overhead included wait event time? Same OS system call for both?
See also this thread from oracle-l:
http://www.freelists.org/post/oracle-l/multiblock-vs-direct-path-read-on-10gR2-aix-related-iostatawr-stat-diff,1
Note that all of these measurements were from 10046 traces rather than AWR or OS reports.
Hello Tanel – can you please post the link to the recorded session? (Sorry if its already on this page – I did not find it).
Thanks,
Naresh
Any repeat / echo sessions ?
.
if you have recorded this session the can you send/post the link of recorded session. In case you have a central repository where you are storing all your sessions the please share the link. Thanks.