oracle database, two tables need to be associated with the query (according to the two fields), and found to not be associated with the acquired count number, given as follows:
Sql same change to the other database can (simply because the amount of data in ten million, it is very slow, when using 40s).
Get it working, consulting co-workers, colleagues, given methods, plus the number of parallel before the count, that is, to change the select count (e.id) as follows:
select /*+ parallel(12) */ count(e.id)
Successful implementation success.
It is said that is because the target oracle database set up in parallel, it is not directly associated with the results of the statistical count queries. No time to carefully read, mark it here first.