Monday, November 14, 2005

Thanks for the feedback...

Thanks for all of the feedback on my Looking for suggestions blog entry. That was fantastic. I chose two topics to attack first. I’m going to do them for the upcoming Hotsos Symposium held in Texas next March. This is one of the few conferences where I look forward to attending sessions as well as giving the session – UKOUG and IOUG are two other examples but the level of technical depth here is something you won’t necessarily find anywhere else. This will be my fourth year there and every year I’ve used that conference as an introduction for new material. Last year it was “SQL Techniques” and “All about Binds”.

“All about Binds” is currently my favorite technical session to deliver. It sounds like a really mundane topic that I’ve beaten totally to death – but it isn’t! Getting into SQL injection and showing it – I am totally amazed that on average less than 10% of the audiences I’ve polled (sometimes as low as 0%!) are familiar with SQL injection. That is scary to me. Demonstrating what this magic thing called “bind variable peeking” is (the part of my session where I “prove” that when it is raining on Monday mornings, you must restart Oracle twice… Not really – I demonstrate how easy it is to fall into false causality without a little research and knowledge…). And closing with the foibles of cursor_sharing = force or similar (and why you really don’t want to use them unless you are forced to – and even then, for as little as possible).

This year, based on inputs from that blog entry (and because they interest me as well!) I’m going to prepare these two new sessions:

  • Performance Impact of the Log Errors Clause in 10gR2: This session will look at the new log errors clause available for use with BULK DML operations in the 10gR2 database. We will see what the clause does, investigate how it does what it does, explain limitations/restrictions inherent with using the clause and finally - compare the performance of slow by slow processing versus bulk DML with log errors - and maybe even compare bulk DML with and without log errors for performance

  • Instrumentation 101: This session will talk about the importance of heavily instrumenting your code and explore the possible methods of instrumenting your code in an Oracle environment - concentrating on developed code both in the database (stored procedures) as well as outside the database (java, C, VB and the like).

These sessions always start as “one hour topics” but over the year will grow into 1 to 2 hour sessions if history is anything to judge by. When I first gave All about Binds – it took 1 hour. Now, I can still do the short version in one hour but really much prefer two hours to do it justice. Same with the SQL Techniques and in fact, most of the sessions. It seems every time I give a session, it gets a couple of minutes longer as I figure out better ways to explain the concept or better demos (based on the questions I get – I refine the material so that question doesn’t have to be asked again – I hope).

I will be referring back to that blog for the other ideas over time so some of the other topics will get included eventually.

On a related note, I was catching up on some blogs and saw this one from Pete_S. It reminded me once again of this blog entry I had a while ago. Design matters, really, it does.


Anonymous Anonymous said....

"restrictions inherit with using the clause "

Mon Nov 14, 11:01:00 AM EST  

Blogger Thomas Kyte said....

didn't get it? did I say something wrong there?

Mon Nov 14, 12:55:00 PM EST  

Blogger shrek said....

OK, so i know at least some of the sessions i'llbe attending at HOTSOS.;-) my first one, so any tips on navagating that conference?

Mon Nov 14, 01:09:00 PM EST  

Anonymous Gabe said....

Just a guess on my part [on that you meant to say and anonymous to point at] ...

"explain limitations/restrictions inherent with using the clause and finally"

Mon Nov 14, 01:09:00 PM EST  

Blogger Peter Lewis said....

I think Anonymous was somewhat obtusely suggesting the word 'inherent' in place of 'inherit'.

I understood you as I'm sure most others did. Down with pedants... after all this is a blog, not a text book!

Keep up the good work,


Mon Nov 14, 01:11:00 PM EST  

Anonymous Patty C. said....

Both session ideas sound great. I am already looking forward to seeing the final product at Hotsos! I can't believe that this will be the 4th year. How time flies.

Mon Nov 14, 01:13:00 PM EST  

Blogger Thomas Kyte said....

I better fix inherent/inherit that before my mom reads it (she was an english teacher way back when and sends me fixes via email :)

Mon Nov 14, 01:21:00 PM EST  

Blogger oracos said....

bulk processing & instrumentation!!! that'd do it! will be there!


Mon Nov 14, 04:15:00 PM EST  

Anonymous Anonymous said....

It scares me that people do not know about SQL Injection. What about SQL Injection attacks? I will say I have "legally" to prove a point killed a server and shut it down using a SQL Injection attack. But the DB was not Oracle. With Oracle it won't take over the server but you can kill the instance, and worse, you can do some serious damage to the data in the Database. I hope everyone has good backups :)

Fri Nov 18, 01:07:00 AM EST  

Anonymous Anonymous said....

Instrumentation 101 sounds really interesting. Any way to get the content without having to come to texas?

Fri Nov 25, 03:47:00 PM EST  

Blogger Thomas Kyte said....

Content will not officially exist until March 2006 ;)

and after that, I'll be reusing it for a while.

Fri Nov 25, 03:50:00 PM EST  

Anonymous Anonymous said....


I would I be correct in saying that your book "expert one-on-one"
is more for developers than DBAs?

Sun Nov 27, 09:05:00 AM EST  

Blogger Thomas Kyte said....

would I be correct in saying

That would be incorrect - it is as much for DBA's as for developers.

Sun Nov 27, 09:11:00 AM EST  

Anonymous getlostdave said....

I attended your Instrumentation 101 at UKOUG. Unsurprisingly good.

One thing you didn't mention about Instrumentation (unless I missed it) was its value in regression testing. Where you can diff instrumented logs pre & post change, you can gain a lot of confidence if your change to your program hasn't produced changes to the log in unexpected areas!

Hope to have the opportunity of seeing you speak again soon!

Tue Nov 21, 02:11:00 PM EST  

Blogger Thomas Kyte said....

getlostdave said...

In 45 minutes, it is impossibly hard to cover everything, but yes, we use that stuff here in Oracle to regression test.

My favorite part of the UKOUG instrumentation talk was when I hit "hibernate" by ACCIDENT with the stupid touch pad. Thought my talk was done - until windows "errored" out and could not hibernate, and restored my session.

That was brilliant, a windows error saved my day :)

How often does that happen?

Tue Nov 21, 03:33:00 PM EST  


<< Home