This week I participated (online option) in HotSOS conference — the only conference that is focused entirely around Oracle database performance and internals. Like a year ago, it was a great conference, I really enjoyed it. Predictably, the number one topic was the in-memory option. Probably every third talk was about it. I learned a lot of stuff I didn’t know before (and I strongly suspect that this stuff wouldn’t be easy to find anywhere else).
In particular, Riyaj Shamsudeen’s talk on Wednesday was entirely devoted to in-memory internals. He shared not just what he learned from long hours of studying dumps, trace files, perf and dtrace output, but most importantly, how he learned it, e.g. trace events that dump information about in memory population and scanning.
Tanel Poder was great as ever — this time he shared a story about troubleshooting a complex performance issue on a RAC environment that had to do with temporary segments created by parallel slaves in CTAS command that cause contention on SEG$ system table, producing a variety of symptoms seemingly unrelated to each other.
I’ve generated (and analyzed) my share of trace files, so I didn’t expect to learn anything new in that area, but Ric Van Dyke proved me wrong with his talk on trace files and instrumentation… really great stuff!
Apart from the “core” stuff (internals, new features, case studies), there was a number of talks on various tools for performance analysis. Carlos Sierra presented his new eDB360 tool which is great for collecting database performance information when access is limited. A couple of superb presentations by Jeff Smith about new SQL Developer features (4.1 “early adopter” edition), and the new command line tool sdsql with great features not available in sqlplus: saving reports as “aliases”, dumping DDL for database objects easily with the “ddl” command, improved “history” (e.g. you can see execution times for all past commands), “sqlformat” that (finally!) makes query output readable (sqlformat ansiconsole) or allows you to generate INSERT scripts (sqlformat insert), and much else. Of course, if you follow Jeff’s blog, or blogs of his colleages (e.g. here), you probably already know about this stuff, but with live demos you also pick little things not available elsewhere.
All in all, it was great. So I’d like to thank organizers and presenters for making this event possible, and my employer for covering the registration fee ($1k well spent!). And if you’re interested in Oracle performance topics, and the next year you find yourself hesitating whether you should go or not, don’t! No matter what level your Oracle knowledge is at, you’ll find something interesting there.
Oracle Diagnostician 