Re: Distributed Join

> relevant to the best solution today. As to distributed optimisation, ------------------------------- > OMNI does optimise distributed joins (because unlike its competitors, --------------------------------------------------------------------- > is has a global dictionary). ---------------------------- Why is a "global dictionary" a requirement for distributed optimization? Since database statistics are typically stored in tables, why not query the remote statistics as you are building the optimization plan? This way... 1) You get the most current data (ie: more accurate statistics) 2) You don't need to constantly keep the "global dictionary" in sync with reality. I would think that a "distributed global dictionary" is a good approach, verses a "central global dictionary", in general. Perhaps if your databases are static, predictable, and few in numbers a central global dictionary is okay. But otherwise...