Monetdb slow query : trace select "aint1p" from "aTestCube"."a1b" where "aint10p"=1234567891 limit 5;

Question

when running this simple query trace select "aint1p" from "aTestCube"."a100m" where "aint10p"=1234567891 limit 5;

I noticed that when running this query on a table that contains 100m rows the mitosis does not work. as opposed to when running trace select "aint1p" from "aTestCube"."a100m" where "aint10p"<>1234567891 limit 5;

why is that? am i doing/configuring something wrong? the nr_threads config is set to 32

Answer 1

the optimizers assume the 'point' query returns only a limited number of rows. A first run of such query should create the hash table and subsequent queries should then be fast, ie. there should be no need for a mitoses partitioning. But the assumption could be wrong...