The past two weeks were nice! I had the opportunity to attend two classes presented by three of the most experienced and talented Oracle specialists in the world. January 19-21 it started with “The Masterclass 2009”, organized by Miracle Benelux and hosting Cary Millsap and Jeff Holt. This was a unique opportunity. I believe it was years and years ago that they visited Europe and they never did a presentation before in The Netherlands.
One of the main reasons for this Masterclass was to honour Lex de Haan.
A week later, on January 25-26, Tom Kyte came to Oracle The Netherlands in De Meern. I think nearly 100 people, mostly Dutch but at least some people from abroad (of which a few familiar ones from the week before) could listen to a lot of information on Oracle 11, storage techniques, rebuilds and good comments on binding.
Millsap and Holt
The Masterclass, perfectly organized at La Place Conference Center in Utrecht, was three days full of a mixture of very technical but also high level information. About 25 people from The Netherlands, Belgium, Germany, Norway, Spain and Poland were attending. Even one from California. Which was nice because he is doing tuning on a 10TB Siebel Database serving 15,000 users. I am only doing tuning on a Siebel system about 1/20th of that size. So I learned something from him, but could also give him some advice that I hope he can use. The most of the class was contributed to problem diagnosis and proper scoping. This is one of the cornerstones of Method R (“Method R Corporation is a software performance company founded by Cary Millsap and Method R is a fast, effective, commonsense approach to optimizing a system” – words stolen from the company’s website).
I have taken interest in Performance Tuning ever since I started using Oracle, but I have been doing development mostly and just since a few years I have the opportunity to do most of my work as a performance and trouble shooting specialist at a big customer site (which I still do). At that moment I started reading everything I could find that was interesting. And off course I read Cary Millsap and Jeff Holt’s “Optimizing Oracle Performance”. And without exaggerating: this book was the one that mostly changed my view on performance, on the workings of computer systems as a whole and on the importance of proper scoping. I suddenly realized why I did not really improve my understanding of Oracle performance after I took the Oracle Performance Tuning course in 1998: this course in those days was all about v$sysstat, v$sesstat, buffer cache hit ratio, latch misses etc. None of this really made me able to diagnose a performance problem and only after reading this book I realized why: it’s all about scoping. Buffer cache hit ratio: not useful. V$sysstat and v$sesstat: forget about it, at least v$sysstat. V$sesstat can be helpfull if you use it as a delta (“What are my stats just before executing the problem transaction, what are the stats directly after it and what are the differences?”). Tom Kyte uses a tool he calls runstats that uses such an approach (using v$mystat though). The last years luckily more people write about new ways of looking at performance tuning. For example I remember the YAPP paper which was really another way of diagnosing than I had seen before.
So being able to meet the writers of this book that was revolutionary to me (and as I heard in the class, to more people) and having three entire days of consuming it again (and more, and told in different ways with a lot of funny stories as well) was a great opportunity. The only downsides were my own condition -which was not the best, especially the first day- and the fact that the timing of the course was not the best since one customer was going into production with a new release and this always means some adhoc time spent looking at last minute problems. But the customer’s laptop and the Miracle Wireless Internet connectivity made that a problem solved, albeit with some extra stress and thus not being able to fully enjoy the class. But hey, that is part of the busy Oracle consultants life ;-).
Without going into details, I will not forget “The story of Bob, Bob’s boy and the Bicycle that became a Ferrari“, Amdahl’s law, Queueing Theory, skew, response time profiles and certainly not the quote from Bobby Bragan (“Say you were standing with one foot in the oven and one foot in an ice bucket. According to the percentage people, you would be perfectly comfortable.”). It was very recognizable, meaning that I did my homework and read the book, but nevertheless of importance to have visited this Masterclass. I did learn a couple of new things but also learned that for some parts of my job (for example my development part) I am actually doing some very good things and am on the right track. But also that I can still improve on that. And that is a confidence booster as well as a goal setter. Above all, I was truly impressed by the tools these guys made in the past and current years. Not only the Method R profiler, which creates a much more rich-full response profile than tkprof does. But also:
- mdist: checking your data for the correct distribution and thus suitability for certain queueing models
- mrls: a special kind of (Unix/Linux) “ls” especially to search and list your Extended SQLTrace files with all sorts of search criteria and format styles.
- mrskew: find skew in response time components in all of the trace files on your disk
- SLA Managerto be able to do SLA’s and check them on Performance of your systems
Incredible and very valuable tools.
Thanks Cary and Jeff for the Masterclass and thanks Miracle Benelux for organizing it! Some photo’s are on their site. And another blogpost can be read on Marco Gralike’s blog, along with his self made pictures. Also check out some other seminars to be organized by Miracle Benelux: Tanel Poder is coming to The Netherlands. I am not sure yet if I will be suited for this class…
The Tom Kyte seminar was different, although I cannot say in which sense. Probably less loose, but there were also a lot more people in the room and the setting was probably less suitable for a relaxed, easy going seminar. But I do not mean it was less interesting. On the contrary: it was two days of new stuff and best practices Tom presented. And he made his jokes! But he had so much to tell, if he would have done the lots of funny stories Cary did, we would have needed two more days.
The presentation slides and all samples used by Tom can be downloaded at his AskTom site. This seminar also showed a lot of familiar material. A lot of it is in his books, the website or in Oracle Magazine columns. But also here very good to hear it live and being able to ask questions to get further clarification. Tom spent a lot of time on the most interesting Oracle 11 new features, on binding, storage techniques and reorganization of objects.
So these were 5 days well spent and with a bag full of best practices I should be able to deliver even better support to my customers. That is what it is all about.