Ariel,
I’ve seen something similar when I’ve added count() to an aggregate query using merge tables. You might try adding an explain statement to see what the server is doing. I am wondering if the server constructs the entire cdr_date column (and starts swapping to disk - hence the 21 min), and then returns the first three entries.
I also ran into problems with columns not being found in the DB catalog of slave instances when using merge tables. It clearly looked like a bug, but I didn’t have the time to put together a reproducible test case.
Dave
From: users-list [mailto:users-list-bounces+david.b.anderson=citi.com@monetdb.org] On Behalf Of Ariel Abadi Sent: Monday, June 27, 2016 2:39 PM To: Communication channel for MonetDB users Subject: MERGE Tables works ?
Hi Everybody!
Im facing a problem. We register telecomunications calls. As the volume is so big, we decided to store the data in different tables. The criteria we used for splitting the info was to create a table per each Switch and Day.
So we have Sw1_20160601, Sw2_20160602...Sw3_20160620..etc etc Each table has aprox 15MM rows.
In order to create some reports we have created a MERGE TABLE and added to this table the 90 tables (3 switches x 30 days). BUT, now Im trying to run a very simple query, like SELECT cdr_date FROM xxxx LIMIT 3; To get at least 3 rows (just to see that this schema works).
It took 21 mins to give me those 3 rows....
Am I doing something wrong?
Rgds Ariel