The First Miracle Open World: Day 2

Oct 1, 2007 / By Alex Gorbachev

Tags: , ,

A ferry seems to be the only place where I can type my blog posts on this trip. It’s 1:27 AM and I’m on the way to Germany now. I had a couple naps earlier today (I should say yesterday), so I feel like getting an earlier start to the drive so I can be in Munich by lunch time.

In this post I’m going to concentrate on the second and last day of MOW. However, a few other events require dedicated coverage that I will need to do later.

The start of the second day was conveniently moved to 9:30 instead of original 9:00. I feel that the organizers really did care very much about the attendees. The Water-Park party was total success and some enthusiasts as usual ended up at the Party House. I plan to cover the Water-Park separately at some point.

Now let’s turn to the “useless stuff”. Anjo Kolk’s and Moans Nogood’s session was swapped with Johannes Djernaes’s presentation about real-time tuning of issues with bind variable peeking. There was a very good reason to shift Anjo’s and Moans’s session to a later time but, first, few words about Johannes’s presentation. I’ve written about bind variable peeking without histograms in the past, and as Johannes mentioned, it somewhat inspired him to do some more research.

Naturally, I was interested in this topic, so I had to give a miss to Jonathan Lewis’s presentation “Understanding Statspack”, which I wanted to see for the third time. This was unfortunate because the 1.5-hour-format session from Jonathan must have been much more interesting compared to standard 45 minutes that I’d seen previously (and it was indeed excellent, according to responses I heard).

There was one more hook in the abstract that grabbed my attention — Johannes promised to demonstrate how to change the execution plan for an open cursor, calling it a real-time performance-tuning technique. As it turned out, the principle behind it was very simple — flushing the shared pool and parsing the statement in a SQL*Plus session with specifically-crafted bind variables so that the execution plan is changed to the better one. I realized that I had actually used this method already few times before, but instead of parsing the statement manually, I let the application do the parse.

I should mention that Johannes used the ANALYZE command to gather statistics. The reason for that was that he found that, even though he specifies “100%”, DBMS_STATS was analyzing indexes with the wrong estimate percent. I admit that I’ve never seen that problem in usable releases of 9i and 10g and always use DBMS_STATS to collect statistics, which I believe is the way to go. Maybe I can check that problem if I have a bit of time.

Back to Anjo and Mogens now. The topic wasn’t confirmed until the beginning of the session, which was billed as “TBD”. I missed the first minute so I don’t know the title, but it turned out they discovered that there are two world problems and two places where we experience them. The problems are skew and latency. (James Morle did a keynote presentation about this at SIOUG 07, but I missed it.) Skew and latency impact two main areas — CPU and I/O, and this is where all performance problems have their roots.

If you have already been to one of Mogens’s or Anjo’s presentations, you know their presentation style — lots of humor and maybe even a touch of sarcasm. Imagine them both together on the stage — double fun! You’ll understand what I’m talking about when I mention a few acronyms they say are generally misunderstood:

  • CRS – Consulting Ready Services
  • SOD – Service Oriented Databases
  • LINUX – Linux Is Not UniX

I should mention that Toon Koppelaars provided an interesting idea. The current trend is to write database-agnostic applications and put business logic inside the application as much as possible. I don’t need to mention that this is a sub-optimal implementation, but what Toon emphasized is that application technologies are much more volatile than databases. Application development frameworks and standards are changing every year and a half or two while databases are very slow to evolve. That means that in two years from now, it will be a problem to find developers that can work with aged-out technology and want to. I was quite amused by that idea. Thanks Toon!

By the way, Toon’s highly-regarded book — Applied Mathematics for Database Professionals — was recently published. He started with Lex de Haan before Lex’s passing. A few people have recommended it to me, so it should be worth a read. I have it in my “to-read” least.

The last session I went to was a masterpiece from Jonathan Lewis — “Optimizing Through Understanding”. It’s amazing how talented one must be to explain the very complex concepts in controlling SQL execution plans in such a way that we mere mortal DBAs can grasp them. Very well done, as usual! I even managed to ask a difficult question! If I hadn’t, Jonathan might have thought that we didn’t pay proper attention.

The closing session was done by Moans and number of others, along with prizes and surprises. I should mention that Cary Millsap ended up in an empty box:

Cary in a box

This box was full of some of the best books about Oracle and databases so substituting them with Cary was very appropriate if you compare how much he knows to the content of the books. I can also mention that Mogens, as well as Jonathan, Tuomas, Cary, Anjo, and Graham, got a CD with 42 songs about Miracle (i.e. with word miracle in them). Carel-Jan Engel solemnly presented IMUG mugs to Jonathan, Tuomas, Cary, Anjo, and Graham. (I hope I didn’t forget anyone).

By the way, I should have told you that the day before, during the Gala Diner, Mogens announced that the IMUG (International Miracle User Group) had been created and that Carel-Jan had been appointed as the Chief of Boards of Directors (or something like that — I can’t recall details as it was a very long dinner, you know). What it’s all about is not clear but you should expect some news about that soon.

I write these finishing lines having already from Munich at my friend’s house. It was a long drive, so I’ll stop for now. When the conference is over, I expect to cover the Water-Park event and the Gala Diner as well as the interesting experience of a Danish sauna. If you have read all the way to these last lines, it must mean that it was interesting. I hope at least a few of you enjoyed this small review.

As usual, stay tuned.

4 Responses to “The First Miracle Open World: Day 2”

Leave a Reply

  • (will not be published)

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>