BIWA Summit - Day 2

The second day has just ended, and now I am back in the hotel thinking of grabbing a bite to eat before the fun tasks of finishing a report for a customer and packing... just hope my stuff still fits my suitcase.

Conferences seem to start early here - and coupled with a 15 minute walk from the hotel to the Oracle Conference Center makes for a long day.

First up was the final keynote of the Summit, Jaun Loaiza speaking on Oracle Exadata. The rest of the day was in technical sessions.  Bryan Wise speaking on Securing OBIEE was well worth listening to. Although I have set clients up with LDAP security before I learnt a couple of useful tips including how to use groups within the LDAP service; in the past I shied against this and preferred to manage groups through a database look-up (and that was still Bryan's preferred method), the idea of writing a database function that uses a call to DBMS_LDAP to return a semicolon separated list of group memberships had not really struck me, he also gave the very sound piece of advice if you are implementing SSO, don't activate it until you have tested your security model, as soon as SSO is on the repository security model is no longer usable - make a mistake and you, the administrator, can get locked out.

One thing I like doing is looking at things I don't see on a day-to-day basis, so the talk on "Crystal Ball" was an interesting diversion for me; then thoughts of my current project beckoned so on to a talk on Master Data Management. Lunch was a 'birds of a feather' type of thing - I started on one of data warehouse tables before going to join one of the OWB tables.
Stayed in the same seat for another, but this time more technical talk on Exadata and the Oracle/HP DW machine, and still did not move for Maria Colgan's  talk 'Oracle 11g Optimizer Uncovered' - three of my big gripes from pre-11g data warehousing seem to have efficient resolutions in 11g; and to think of the hoops I had to go to get around them in Oracle 9.2 or 10g.  Getting cardinalities right for correlated columns, keeping optimal query plans after patches / upgrades and quickly maintaining global stats on partitioned tables have been a major issue for me when I used to operate (and not just develop) data warehouses. Her other feature, bind peeking on skewed columns is less of a problem for me as mostly I don't see bind variables in use.

Finished a long day with talks on SOA and unstructured data.   All-in-all a great conference, I am already looking forward to the next one!