The Travelling Man

Like a few other people I was up in Edinburgh earlier this week for the OUG Scottish SIG. I was asked to present at a fairly late date on Business Intelligence 10g, and as the short BI presentation I did at the last Scottish conference went down well, I thought it was worth putting something together. A bit of a departure for me this time was the fact that I did a live demo, rather than just rely on slides - a preference for demos came out fairly strongly at our last UKOUG BIRT SIG, which is fairly understandable if you're talking about new client tools (which obviously are fairly visual in nature), so I put a short demo together that showed how the new Discoverer tools work together and integrate into Portal. If you're interested in looking at the slides I've uploaded them here and if you're looking for screenshots of the various tools there's a (alternative) longer version available here.

Talking of the conference it looks like I'll be doing a couple of presentations on BI 10g and the OLAP Option. The first one is an update on Oracle Business Intelligence 10g, and looks at what's coming with the Phase 2 release due in the second half of 2005. By that point Oracle Warehouse Builder "Paris" will be available and I'll take a  look at how Paris can be used to build a OLAP-enabled BI system that uses Discoverer, BI Beans and Portal as the front-end. The second presentation focuses in on the OLAP Option and Discoverer for OLAP, and looks firstly at how cubes are created and then goes through some of the Discoverer for OLAP-specific techniques and features that you can use with OLAP data. The point of the first presentation is to give everyone an update on Oracle BI in general, and show all the tools off in an integrated demo, whilst the second talk takes a more in-depth look at the use of OLAP and tries to show what extra you get when you use the OLAP Option.

Back to the Scottish event for a second, probably the best bit for me was the chance to meet up with some of the people I've got to know via the blog over the last few months. Doug Burns, Niall Litchfield and Mogens N rgaard were all speaking, and we all had a chance to meet up (along with Thomas the SIG deputy chair and Peter Robson, one of the UKOUG directors) in one of the pubs in the centre of Edinburgh. Doug in particular had eyebrows even more impressive than mine (see the photo hosted on Doug's site) and Mogens was as interesting as I thought he'd be (Niall's pretty cool as well). It's making me think actually that it might we worth trying to organise a get together for one evening during the UKOUG conference, possibly on the Monday night, open to anyone who participates in the Oracle blog community (writer, commenter or lurker) - drop me a line if you think it's a good idea. Mogens' presentations were pretty good as well - I missed the first one as my presentation clashed with it but got to hear about it the night before, but the second one, that looked at indexes (and the fact that they don't scale, which is why you never see them used in the TPC benchmarks, with hash clusters used instead) was in particular an eye-opener. If you're a UKOUG member you can download the presentations from the event website, otherwise I guess ask Mogens nicely and he might send you a copy.

This week (and the week before it, and the week to come) have been pretty busy with me having to travel up to Edinburgh, down to Reading, up in Manchester last week and off to Liverpool (and then Reading again) next week. It's been a bit of a week of contrasts (accomodation-wise) this week, staying at the Malmaison Hotel in Edinburgh at the start of the week (ultra-trendy boutique hotel in the arty part of the city), followed by a night in the Reading Travelodge on Wednesday. A tip to anyone looking to stay the night in Reading - don't stay at the Reading Travelodge, unless you're wanting to do yourself in and want something to urge you on to finish the job off. A converted office block overlooking a busy road junction and with facilities that consist of a vending machine (and that's it) I spend a happy five hours in my room with just a bed, a table, a television and a small bar of soap. At least I had no excuse to get a late night though but I think I'll probably be looking elsewhere next time I stay down there.

I mentioned earlier on today that my new article for OTN has just gone up on their website, and if Discoverer and the OLAP option is something you find interesting, you'll be interested to know that I've just agreed to put a chapter together on Discoverer for OLAP for a new book due out later this year. I'm going to take a good look at the OLAP version of Discoverer, what it's used for (in relation to regular Discoverer), how it works and how you can put together queries that take advantage of the unique features in the 9i and 10g OLAP engine. This will be my first published work (if you don't count the various user group magazines) so I'll probably have my head down and be out of communication for the next few weeks, while I put it together. Once it starts to take shape I'll be able to put our more details, but until then if there's anything about Discoverer for OLAP that you think it's important I cover, drop me a line.

Just one last bit of Oracle news. Looks like Larry's had his cheque book out again as Oracle have just purchases TimesTen (news.com, The Register.) Unlike some of the recent Oracle purchases - Retek, Peoplesoft, Oblix - this one looks like it will have implications for the core RDBMS product, as TimesTen seem to be all about in-memory databases used for high-end transactional and analytics applications. From what I can make out, this is more to do with specialised small databases that perform lots of transactions quickly for customers in the financial services, telecoms and so on industries, rather than the idea of caching an entire multi-gigabyte SGA in memory using RAM disks. Still, it'll be interesting to see how this technology folds into the core database product - maybe it'll become an option, like the OLAP option, with the ability to cache transactions in the "TimesTen" engine and access the data faster - or maybe the technology will be used to enhance the current way that Oracle caches data in buffers. We'll have to wait and see.