Larry Ellison will announce that Oracle 12c can now work "in-memory". Can it be as good as a purpose-built in-memory database ?
The very first SQL dialect that I learned was Oracle - way back in 1989. In the 25 years since then, I've worked with a dozen other dialects and Oracle has grown considerably.
I don't just mean "grown" as in "makes more money" but also that the core product has spread out to cover a lot more ground. There have been all manner of innovations and acquisitions that have brought the Oracle database to its current position where it seems that there is nothing it can't do.
And in a few hours Larry Ellison will be announcing the details of how it can now also effortlessly (?) work in-memory "at the touch of a button".
The question for me is not "Can it do everything ?" but "Can it do everything well ?"
You see, I'm a fan of buying a tool for a specific job. I could for example cut the hedge in my garden with my lawnmower, if my arms were strong enough. Instead I bought a hedge trimmer for that job.
In the same way, I could buy a whole bunch of attachments so that I could rake, pick up leaves, carry cargo, spray water and so on with my lawnmower. Just because I could do this doesn't mean that I should. A lawnmower that can do everything probably can't do the non-cutting jobs as well as a specialist tool could and you would worry that you might get a better cut lawn if you used a lawnmower from a company that doesn't compromise the design and (shock, horror) just concentrates on cutting grass really well.
That's my preconception of what Oracle's in-memory will look like - another attachment for a product that is already dangerously overloaded with attachments and which won't cut the hedge as well as a hedge cutter could.
Larry Ellison is always an entertaining speaker though, so I shall definitely be listening in, even though it will be 19:00 German time and I probably should be home cutting my grass ...