OUG Scotland

Well, to start with, it was a great venue. Perfect size, perfectly ‘Edinburgh’ and a nice coffee/bar area. I understand some people were freezing cold, but I think I got lucky. It was also nice to see quite a few familiar faces, including the entire Edinburgh University IT posse! (At least you can get a decent chat about football with them, too)

First up was Geoff Locke from Cap Gemini with a “Fast track to using Oracle’s Wait Interface (OWI) for Tuning”. This was probably my least favourite presentation of the day because I didn’t feel he spent much time talking about the wait interface so that was a bit disappointing. He had clearly based his presentation on his work in the field, though, which is probably more practically useful than me waffling on about px tests, so it depends what you’re looking for.

Next was Simon Kelsey from BSkyB (one of my colleagues) with his very first user group presentation on compression and some of the performance implications. He did a great job and I wouldn’t have noticed he was nervous if I hadn’t known already 😉 He seemed to enjoy himself too. As a Geordie lass might once have said, ‘if he can do it, anyone can’ 😉

Jonathan Lewis gave a couple of CBO presentations during the day but I particularly enjoyed the so-called ‘introduction’ in the morning. Where Jonathan says ‘Introduction’, the rest of us hold our breath and listen carefully! To pick out one point which drew a few gasps from the audience, he illustrated how dropping an index which is not even being used by a query can change the optimiser’s strategy for the selectivity calculation and cause it to choose a different plan. I’m sure Peter Robson described this presentation as a ‘tour de force’, but I wasn’t really listening to him. I must stop doing that or I’ll get caught out one of these days!

After lunch Moans Longballs Nogood reduced the room to fits of laughter, as usual, and also had some pretty interesting things to say about Opterons, operating systems and old men. He’s an expert on at least one of them 😉 I spoke to a few people who had never heard Mogens speak and they were bowled over and charmed as usual. If you ever get the chance to see him, grasp it and listen carefully. He’s so busy taking the mickey out of everything and everyone that you need to pay attention for the good bits and make allowances for his martini-dry sense of humour!

Of course, he overran, leaving less time for some bearded bloke to talk about the optimiser some more but that was okay, because he just took his full time too. Not content with that, he filled the audience’s head with even more startling information. Good job they had something light to round off the day (and a nice convenient coffee break before it for people to slip out the door early!)


I picked this photo out of the several that Peter took for me because Mogens thinks I look like I’m about to break into my Rock and Roll act! (Don’t stand in front of the projector Doug!) That’s Thomas Presslie doing the introduction.

I think this was my least favourite delivery of this presentation but it went okay I think. It was the end of a long day, everyone was a bit tired and I was keen not to over-run or people would kill me. Hopefully it got people thinking about the possibilities for their own testing and some might read the paper. I’d be happy with that result.

Of the couple of things I didn’t mention because of time constraints, one was that I received an email on my phone during the day from Jeff Moss about some of his results with reducing the degree of parallelism on his system which look quite promising. He’d also had a very good tuning result on a problem query which I helped him solve. What was the extent of my expert assistance? I suggested that he might want to read the Hash Joins chapter of Jonathan’s book. He did, and found the answer. I meant to mention this to Jonathan when having a few drinks afterwards but the conversation was so intense and wide-ranging that it slipped my mind.

Good day, great organisation and how nice it was to be able to walk home through Edinburgh on a sunny evening (with my gift of a bottle of good single malt whisky) instead of sitting around in an airport waiting to be called.

7 comments

  1. If y’all look carefully, you can see the microphone in Doug’s right hand. His feet are wide apart. Look at his determined face. He’s ready to do his rock’n’roll act. And there’s the word “UKOUG” across his flat, hard belly.

    This man is ready.

  2. Oh, I nearly forgot: I’ve got a video of Thomas Presslie – the guy who makes these splendid events possible – accessing his own house through an open window because his wife had locked the door while we were at the local pub. The video has some explicit footage, but it might be OK to show it to DBA’s. Oracle DBA’s, that is. Not SQL Server DBA’s.

  3. The expert assistance from Jonathan (via Doug of course!) was that I had misunderstood which way around the hash join tables should be present…if you use an ORDERED hint that is…I had them the wrong way around and it seemed to improve the elapsed time (a little) and the memory usage (quite significantly) when I reversed them (p320 ch12 CBO Fundamentals) – I still don’t understand all the logic behind it because, as ever, Jonathan’s book is far more detailed than my wee brain can absorb in one session.

    Still, as you point out Doug, more and more testing of this px stuff is pointing to lower DOP’s than higher ones…unless you basically own the box with your single query/dml/ddl.

Leave a comment

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