UKOUG Part Deux
Posted by John Hallas on December 3, 2008
Yesterday I finished my blog with a mention that I was going to see Alex Gorbachev talk. Well I did and I didn’t. I saw the clusterware presentation he had written, he was on the podium but he did not do the presentation. It had looked like he might not get to Birmingham on time so he had called a fellow Pythian in at short notice. The stand-in did a good job but it was quite obvious he did not know the material or presentation and the comments I heard later all aligned with mine. Alex should have taken over. Saying that there were at least 2 slides that were very useful. One on the names and location and order to study event logs if problems occured and the other relating to which processes call other processes.
Wednesday morning started with Understanding Lobs by Julian Dyke. I was looking forward to this and whilst it was detailed and pretty complex with lots of block dumps I am not sure it answered the question that Julian posed early on. He was saying that he had become aware that a lot of sites were making basic mistakes with their use of lobs. I was hoping he might provide a summary of what those mistakes were (and the fixes) but whilst he might have alluded to the problems I don’t think he truly covered them all. However I now know much more about lobs and sizing chunks and parts of the talk about compression factors is one that I have marked down to study further.
Chris Dunscombe did a talk on SQL Tuning which I enjoyed and was very informative. On the way out I heard a couple of people saying that they could have provided a few examples of badly written SQL and how they had fixed them. Yes, very, true ….but they didn’t did they.
Dan Morgan (of PSOUG fame) did a talk about how to manhandle Oracle into hanging, crashing, corruptions, spins and anything else bad you can think of. The aim was to learn how to handle the resultant issues. I think most of us had tried a block recovery using RMAN but probably not that many of the other tests.
Niall Litchfield provided the next seminar which was a sort of travel back in time through Oracle history showing how you should use the tools on offer at the time and free yourself up to do more specific or targeted work. Forty five minutes of easy listening to a well-written presentation that was not too challenging.
Stephen Haisley from Oracle talked about Optimising Oracle Backup and Recovery Operations and I picked a few tips up from that. My final session of the day was by Steve Shaw (I think he might have written HammerOra, or at least had a lot to do with it). He was demonstrating its use for large scale parallel query testing. He presented very well, knew his subject and provided lots of good slides.
So that was the end of my three days at UKOUG. I enjoyed it very much. I am not sure I learnt vast amounts but I have got a lot of notes about things to check out and from there I will undoubtedly get more knowledge, which is the way it should be.
The organisation was superb. Every presentation I saw had been well written and prepared and the speakers were universally good. I might have been slightly negative with some comments above but there is no point me stating that everything was 100% perfect and after all it is only my view.
I think the main thing that struck me is those who are well known (and therefore experienced) on the speaking circuit managed to talk with an easy familiarity that seemed to make the whole session seem so much more enjoyable.