Hotsos 2011 – Day 1

Posted in: Technical Track

Yesterday was actually my second day at Hotsos, 2011. I arrived at Dallas late on Sunday night, and had a bit of time to catch up with friends and colleagues, but not much else.

Monday the presentations began, and early in the morning at that.

Kerry Osborne gave the keynote – some kind of “History of Computing” thing. The subject was slightly less than fascinating, but Kerry is a funny guy and it was a fun keynote anyways.

The next session was also Kerry, this time speaking about Exadata performance. I can’t say I learned anything new. This is usually an issue when attending a presentation about hot new topics – too much time is spent on the basics. In the past every presentation about streams performance included 30 minutes of “what are streams” slides which bored me to death.

Then came Cary Millsap with “Thinking Clearly about Performance”. It is really lovely to watch how his presentation evolved over the year or two that Cary was presenting on this topic. Cary is a master presenter – He has clear and minimalistic slides that support his talk, he has great stories and awesome delivery.

I appreciate the need to think clearly, so after lunch I attended Toon’s “Thinking Clearly about SQL”. This was a thought provoking presentation – I stayed up late last night trying to figure out what I learned there and what I think of it. Toon Koppelaars advocated formal methods of writing SQL statements. This involves two steps – converting business requirements to a format problem statement and then converting these statements to SQL code using formal transformations. The first step seems mandatory to me, however the formal conversion has high overhead and may result in very clunky SQL. I’d only try that if every other method of arriving at correct SQL will fail.

The next session was Andrew Zittelli’s “Four Things Every Developer Should Know about Oracle”. It was excellent. I want to send the slides to every developer I ever worked with. I learned quiet a bit about how redo works for parallel sessions and some about dbms_locks and driving tables.

My own session was up next, which caused me to miss “Contemporary Latch Internals” by Andrey Nikolaev – everyone says it was excellent and that I should have been there. My session went well – the attendees had great questions and for once they even laughed at my jokes. However, I believe that my TCP/IP presentation is cursed to always finish 10-15 minutes early. It happened yesterday, even though I’ve added a 10 minutes section about RAC troubleshooting to beef it up. For Collaborate, I’ll practice speaking slowly.

The last session for the day was Golden Gate performance by Stephen Haisley. It  was a good session and pretty in-depth. I learned some important things about monitoring GoldenGate, how to use the latest version (and when to avoid it) and why its important to use SQL batching.

It was a good and full day, and it continued with fun discussions over dinner and a surprisingly technical talk late into the night in the local bar. Hint to the wise – spending an hour or two in a bar with someone like Christo or Dr. Neil Gunther, and you learn more than you do in a full day of conference.



Want to talk with an expert? Schedule a call with our team to get the conversation started.

7 Comments. Leave new

Did you talk about faxing this time? :p

“Hint to the wise – spending an hour or two in a bar with someone like Christo or Dr. Neil Gunther, and you learn more than you do in a full day of conference”

That’s my secret sauce. I corner, I mean buy drinks for, super smart people, sit back and listen (learn).

Alex Gorbachev
March 10, 2011 4:06 pm

Faxes have been mentioned on private setting…

Gwen Shapira
March 10, 2011 7:47 pm

I am lucky – super smart people usually buy me a drink.


Hey Gwen, why would you avoid the latest version of GoldenGate?

Gwen Shapira
April 8, 2011 2:09 pm

Hey Stephan,

I have to admit that I’ve no idea what I was trying to say there, but I’m sure I didn’t mean that upgrading to GG 11 is a bad idea. A lot of the enhancements sound important and useful.

I *think* I meant that you mentioned a process that could be configured to work with a new 11g method or keep running the old way and there are pros and cons to each decision. On the other hand, maybe I hallucinated, it was the last presentation in the day after all :)


Yes, the Extract can be configured to use the new ASM log reading API (preferred for higher performance) or the older one that uses a connection to the ASM instance and limited due to using a PL/SQL interface. The advantage of using the older one is it can be used when the source database is not open, which the new API requires. But it is 3-4x slower if your redo generation is high enough.
I hope that helps.


It does help. Thanks.


Leave a Reply

Your email address will not be published. Required fields are marked *