Oracle OpenWorld Day 2 Highlights

What a difference a good night’s sleep makes to one’s outlook on life !! Slept soundly until dawn for the first time since arriving to the US and boy, do I feel better for it. After a lazy morning, had a late breakfast at my favourite haunt, Louis Diner, for a breakfast that lasts me all day.

The first session of the day was a presentation on SOA and the so-called transformation the IT landscape. I’m very curious to see how other organisations implement SOA solutions as I have reservations regarding how my work is contemplating going down this path. It’s was actually quite a useful summary of the various integration issues that many organisations have and how a SOA based solution attempts to address these issues. It was a tad marketing oriented by those involved but worth sitting in on.

More than can be said for my second session on Oracle Database Performance and the latest developments, a boring marketing driven piece on how Oracle can scale this way and that, how it’s got the best results here and there and how it’s the market leader here, there and everywhere. I joined the every growing stream of people walking out and decided instead to check out the exhibition hall. As usual, the place was huge and packed with vendors desperate to get your details in exchange for a t-shirt or flashing yo-yo. I actually had a good chat with a number of vendors and saw various database monitoring products that was worth further investigation. Caught up with Gary Goodman from at the Hotsos stand, had a chat with a lady who was an ex-watering skiing champion who loved competing in Australia and of course left with a bag full of t-shirts and flashing yo-yos.

Next was my second official presentation on index related new features introduced in 11g. There was a good turn out with few falling asleep so hopefully it was reasonably successful. I was a bit dehydrated with my voice threatening to give out but I manged to make it and finish in time no less. Both Asif Momen and Jarneil both appeared to have enjoyed it.

I next attended a customer panel on Oracle Exadata and IT consolidation as it’s a technology that is obviously very topical at the moment. It’s of course another marketing exercise however if their joint experiences are to be believed (eg: implementing 1/4 rack systems with massive performance improvements and heaps of free resources to spare), then it’s potentially a technology to consider moving forward. The devil of course is always in the details however and it’s something I would love to test with our production workloads. I can’t help thinking Oracle’s move towards larger more powerful node Exadata systems, with fewer actual RAC nodes, suggests an interesting direction, especially for OLTP based systems. In all, a useful session with plenty of food for thought.

I had just enough time to pop over to the OTN Mason Street tent for a quick drink and bite to eat to attend a “Meet An Oracle ACE” function that was being hosted. Met a few new ACEs myself, including a Java Champion who was surprisingly very normal and nice 😉

My final session of the day was the Real-World Performance Panel session, which is always worth attending. I managed to catch up with Graham Wood and Greg Rahn beforehand which was nice. This year there were actually quite a number of questions that were related to indexes (perhaps someone noticed I was sitting in the back row !!). The question that most grabbed my interest was whether indexes are required at all in an Exadata systems or whether everything runs so fast and vast volumes of data can be processed so quickly, that using indexes is all a bit redundant and unnecessary. The general answer and consensus of the panel was that it of course depends. The fundamentals with Exadata are just the same, it’s just that Full Table Scans and parallel processing can all be done so much more efficiently and quicker than previously, resulting in FTS and hash joins being more likely and attractive to the CBO. However, for the policing of constraints, for row (or “few” or relatively selective rows) processing, then accessing the data via an index is still going to be a more efficient option. It all still depends. For Data Warehousing however, thinking in terms of set and parallel processing should certainly be considered which in turn naturally leads to less reliance on indexes. However, it again all still depends and is just as relevant in non Exadata data warehouses. In all, it was an excellent session with some good questions and even better answers.

After a quick rest back at the hotel, it was time for the Asia-Pacific cocktail reception at the rather impressive Press Club where both Chris Muir and I managed to get on good terms with a number of the waiters who made sure we were served first with any new food being handed out. The  combination of chocolate truffles and red wine was just divine !! Finally, we made an appearance at the massive OTN party where we met up with a number of folks over a wine or two. One thing that can alway be said about the OTN team, they sure know how to organise and throw a great party.

Overall a mixed day with some disappointing sessions, some really good ones and with a number of new t-shirts in my collection 🙂