> Interesting. Their website is full of marketing speak, > but very light on what makes this database so "revolutionary".
Same impression here. Seems like they have built a clustered DB engine from scratch. That's good for throwing out the legacy cruft that all RDBMSes stagger under, but bad for maturing the query optimizer and the management tools. Possibly they're also doing something new on the clustering front, along the lines of decentralizing query and storage optimization. Case in point, the current beta lists as one of its achievements: "The system now uses heuristics to intelligently pick optimal index." Now the optimization fun begins for them :-) And this one: "More complete support for the top ORM tools, such as Hibernate for Java and Active Record for Ruby on Rails." Which I translate to "they're still ramping up the metadata". Beta 7: "Up to 50 nodes". Ouch. So much for scalability. Or they are still working on fine tuning, and 50 nodes is what they've got covered. Hm. Actually I'd say that 50 nodes is beyond what you're do out-of-the-box already. Today's clusters start to require very much special attention beyond a dozen or so machines. 50 EC2 nodes (i.e. no hand tuning for intra-cluster communication with 10-gigabit ethernet and such) is pretty large already. They say they're currently backed by 10 millions of investment money. That's proof-of-concept stage, and a candidate to burn experimentation time/money on. It would be interesting to hear who's backing Malcolm's efforts here: Nuo, Apache Foundation, Malcom himself out of personal interest, somebody else?
