UKOUG Days Three and Four

I was up pretty early on Wednesday to carry on working on my presentation. As I said to someone, once they saw how basic the content was, it would be difficult to see why it takes so long but a lot of it is tweaking with the flow and by going over it so many times in the final days, it really helps it stick in my mind.

But nothing was going to stop me hearing Graham Wood talk about the "DB Time Performance Method", an alternative to other methods such as YAPP and Method R. As regular readers will know, Graham is an architect at Oracle with deep involvement in the ASH, AWR and ADDM features. He talked about performance tuning in general, existing methods and how DB Time could be used instead to similar effect. He showed the various places where DB Time is exposed (and yes, he was asked a licence-related question again!) and how they could be used. The benefit of these features is that they build on existing methods but may be applicable in more situations, for example when a problem is difficult to recreate, because data collection is always enabled.

At one point he said that he'd like to come up with a better name than the DB Time Performance Method, so if anyone has any suggestions ....

I bumped into Marco after his presentation (it sounded like it went pretty well) but by heading along to the speaker lounge with him for a coffee, I got caught up in quite a few in-depth conversations and ended up being pretty late for the start of James Morle's "Every Performance Problem Is One Of Two Things".

Those two things being Latency and Skew. I particularly liked the treatment of latency and I can only reinforce James' message that a small increase in latency can lead to a big difference in performance if it's multiplied. This is very similar to an incident I came across a while back, involving an application design based on a very chatty app server, leading to the same wait events that James discussed. As usual, he was entertaining and funny, too, which is always a help when you're attending a few presentations in a week. Good stuff.

Then I knew it was time to go and hide in my hotel room to attend to my presentation and blog maintenance! I find it very difficult to work on a presentation at the conference site because there are so many people to talk to instead By now it was coming along pretty well, but I knew that I wouldn't be doing much more on Wednesday. I attended a tiny bit of one presentation (and thanks for the support, guys) and then went to meet the Belgians (again!) and Daniel Fink. Yet again, it was 5pm and I hadn't eat a thing, so good on the Belgians for suggesting some food!

After that, it was back to my room for some more prep, a quick snooze and then I realised that I hadn't spent as much time with people as I would have liked (I only get to see some once a year), so I accepted an invitation to The Tap and Spile, which was packed to the gunnels with conference attendees The list of Oracle faces in that pub would warrant another blog. However, I did make a point of thanking some of the CERN guys for attending. As well as the presentation I managed to attend, I heard excellent reviews of several others. These guys are just the type of top-quality contributors the User Group needs and very modest and good fun with it. I hope they become regular visitors.

I had a couple of pints, made my excuses and left at 11:30 so I could be up early. However, I hear that some stayed there until 4:30! Now, I *know* I'm getting old, because I could never manage that! (Then again, I saw the state of them in the morning )

The final day dawned and all I really managed was my own presentation.

It was one of those rare occasions when I felt good about it at the end. It *was* very light-weight technically, but I think it flowed quite well and the audience reaction was excellent. I think sometimes you need a little opinion in the presentation for people to connect with, as well as a little technical content.

Updated Later

Thanks (I think!) to Peter Robson for the following photo. I can assure you I didn't spend the *entire* presentation with that look on my face! (and, *yes* I *do* have a red laptop)



I was very pleased to see Graham Wood in the audience but hadn't realised how much he had worked on OFA, as well as Cary Millsap and other guys at Oracle. That was useful, though, because he was able to clarify one or two points and afterwards we had a terrific in-depth conversation about some of the more recent history of OFA. That was one of the highlights of my week. (Oh, and Graham, it ws Nigel Thomas who asked me to say hello!)

In fact, I had quite a few conversations with people after the presentation which is always a big part of the conference. (I was particularly pleased to see someone who attended my presentation had already downloaded Cary's OFA paper to his PDA to read on the trip home - mission accomplished!) Unfortunately, it also meant that I had a toss-up between lunch (hadn't eaten again) and Jonathan Lewis' Masterclass. The stomach won. Then it was off to the airport and back home.

I wish I'd had time for more fun and learning during the conference. In future, although I know I can work on a presentation during a conference (useful if life has been busy in the lead-up), it's definitely not the best approach. A few tweaks, maybe, but not the entire presentation

It was still a good week, though.