Disclaimer: I work for ScaleBase , I live and suffocate every day ...
I would advise here that if you outline, for example, in column A, then WHERE with columnA = xx will go to one shrad. WHERE columnB = xx must go through all the fragments, because in all of them there may be a column B = xx. If columns A and columnB are not specified. And then you really need to keep the relation in the mapping table. I can say that work on all databases can be super-fast, you need to run in parallel and combine the results. In ScaleBase, we support combining ORDER BY, GROUP BY, etc. It's not easy...
Hey, see more details on your blog: http://database-scalability.blogspot.com
source
share