In the morning, I saw fellow ACE Director Eric Marcoux present an overview of all Oracle Portal Products – an ambitions undertaking, given that Oracle has four portals and Eric also covered Universal Content Management. With such a wide topic, it is probably unavoidable that some errors creep in – for example, I noticed that he erroneously claimed that Oracle Portal does not support JSR-168 and WSRP, which since 10.1.4 it does. But while I feel he showed Oracle Portal in too negative a light, there can be no doubt that Oracle WebCenter is the strategic product.
In the Unconference, I heard Bex Huff (another ACE Director) present on Enterprise 2.0. His sobering observation was that most Enterprise 2.0 initiatives fail (at least initially), and the failures are more likely to be cultural than technical. But that’s supposedly OK, because nobody know how to build an Enterprise 2.0, so you’ll have to accept some failures along the way.
In the Middleware General Session, Thomas Kurian once more announced the imminent release of JDeveloper 11g. One interesting focus area of JDev11g is Application Lifecycle Management, which includes integration with all kinds of third-party tools for version control, build, bug tracking etc.
Oracle is into process modeling (again – remember Oracle Designer?) with both Business Process Analysis and Business Process Management. While I understand where BPEL fits in, the distribution of work between BPA and BPM is unclear. But Oracle claims round-trip engineering between BPA/BPM and BPEL, so you can supposedly let an analyst draw a flowchart and give it to the BPEL developer, at least as a starting point.
With JRockit (a Java VM that doesn’t freeze every once in a while to do garbage collection) and Oracle Coherence (middle-tier long-term object cache), it seems that Oracle has now purchased the technology for some very high throughput middleware architectures. The performance metrics they presented on fairly modest hardware were impressive.
In the Database General Session, Andy Mendelsohn went through all the reasons people should move to 11g – for example you can now use compression on everything, sometimes even getting a performance benefit, because you need to read fewer blocks from disk. There was also the fact that you can now read and write files as fast to/from the database as from a file system – amazing! And there’s a graphical Explain plan in Grid Control (coming to DB Manager), data modeling and E-R diagrams in SQL Developer (another point were we’re getting what we had in Designer…), and a wizard in Application Express that you can feed a Forms XML file and let it build an ApEx application with similar functionality.
It has also become much easier to upgrade. If you have the luxury of a separate environment, you can record your actual workload and play it back on the upgraded environment to see how it performs. And for those (most people), who do upgrade-in-place, you can set the wonderful parameter OPTIMIZER_FEATURES_ENABLE to your current version. This means that your 11g database will use the same execution plans as the old one would. You can then set OPTIMIZER_CAPTURE_SQL_PLAN_BASELINE to capture how your SQL is being executed. Once you have a baseline, you can then keep the plans from the baseline, but let the Oracle optimizer store all the alternatives it comes up with. A DBA can then review the suggested new plans and decide whether to implement them. Cool.