While attending the 11gR2 launch event in Milan last Thursday, I had the distinguished opportunity (invited, as a blogger, by the Oracle team that was organizing the event) to meet Mark Townsend and exchange a few words about the new features of 11gR2 and the Oracle database in general as well.
For those who don't know, Mark is (among other things) the Vice President in charge of coordinating the Product Managers and a technical expert at the same time, and this rare combination has the advantage that you can ask him about any feature you like at whatever granularity you like, from the strategic level down to the technical gory details. In fact Mark is frequently seen at public events (such as Oracle Open World), speaking to mixed audiences composed of both Engineers and Management.
We spoke about many new features of 11gR2, so much in fact that in order to do justice to the information that Mark very kindly provided me, I will use them as the foundation of some future blog posts. In brief anyway, I noticed that this release has features that are targeted mostly to vastly improve the "grid", but with very interesting features for the "core" as well (my personal favourites being the in-memory parallel execution and the new SCN-based MV log, which I plan to blog about in the very near future).
But especially, I didn't miss the unique opportunity to discuss with one of the top players of Oracle Corporation about the amount of information that Oracle shares with its professionals, obviously trying to push for much more. I'm sure that everyone that works with any kind of software product agrees with me on the fact that knowing how the product works is key not only to troubleshooting (that being almost obvious) but to good design also, or perhaps especially; the more you know, the better designer you are going to be.
We had a lot of back and forth on this topic, but to summarize, Mark agreed on detailed information being very useful for experts, but also pointed out that it takes years and years to become proficient enough to be able to digest very detailed documentation, and in the meanwhile, too much information is going to confuse, rather than clear things (especially for juniors coming from other database products). Just the current detail level is overwhelming, the current docs being composed of a staggering 21,000,000 words - and Oracle is in fact trying to organize the documentation in layers as much as possible, starting from high-level descriptions (the Two Day DBA course), than the Concept manual, and than the rest. The most details, however, will be still provided forever as Metalink (aka "My Oracle Support") notes, the only reason being to avoid confusing people, not to hide anything (besides strategic algorithms not covered by Patents of course).
So in short, from this discussion I have understood that Oracle is willing to share information with its community; it only wants to find the right way to do so, since the community is huge (hundreds of thousands of people) and the product is very, very complex. Anyway, I have insisted for more information about the two topics that I'm sure that are not documented well enough, naming the "auto" features and the CBO algorithms - and actually I feel like I have insisted perhaps too much on that with Mark ... but that's something I really care about, both for professional reasons and keen interest alone.
Well, I might add that information sharing, and community involvement as well, is in my opinion one of the main factors of Oracle's success; actually, being able to dig a lot into the inner workings of the product is the reason why I chose an Oracle career ten years ago.
PS I'm back from my vacation and I have a lot of interesting things to investigate at work that look like perfect candidates for being turned into posts, so I will be able to blog more frequently in the future. I also have a series of posts about the CBO that is "almost complete".