Hi,
MonetDB does not support user defined clustered indeces. To have the same effect you will have to sort your data on x-y be explicitly stating it on your insert command, like:
INSERT INTO table_name (...) ORDER BY X,Y
I guess it is sufficient to do this when inserting the last tuple? I am loading up to 25M tuples, so sorting after each step is certainly a bad idea ;)
MonetDb then will know that your data is sorted and result in optimized plans. As far as I know there is no command to see the indices since there are no user indices.
b.) Does MonetDB support MergeJoins? I.e., assume I have a clustered index on columns R1.a and R2.a and perform a join on these columns, will this join be realized by an efficient merge join?
If your data is sorted, MonetDB will choose a fast merge join.
OK, I will try the approach you proposed above, and compare the execution plans.
Which version of MonetDB are you using? Stable release or from the cvs?
Just installed the CVS version from today. And maybe it is of interest that I am running the queries with the new algebra (flag "-G"), which might affect the execution plan.
Will report on my results later, thank you very much.
Michael