比连接更好的解决方案

Question

I have a mySQL query with a join on 11 tables with approximately 50000 rows in each table, all of them linked by a serial number common in all tables
Although i only want one row per serialnumber, but mySQL takes ages to run the query as it has to join so many tables with so many rows
i believe it has to create around 50000^11 (11 in power)

Example: (the sample query may not be syntactically correct but i hope you get the idea)

select distinct serial, one.data1,two.data2,three.data3 from list 
left join (select * from table1)  as one on list.serial=one.serial 
left join (select * from table2)  as two on list.serial=two.serial 
left join (select * from table3)  as three on list.serial=three.serial 

are joins better or sub queries?
is there a better way to build the sql query which takes less time?
if yes, how?

Answer 1

One solution is to create and maintain a separate index table that consists of the datapoints that you need (serial, one.data1, two.data2, three.data3 for example). Reindexing (the query and insertion of records) this table will result in overhead, but all subsequent queries can save a lot of overhead.

Another solution that I have found useful is to perform programatic joins. When multi-table joins with simple criteria require too much memory, there can be a massive performance degradation. You should try benchmarking manually joining datasets built from easy queries using whatever native programming language you are working in (PHP for example). Sometimes you might get an edge using this technique

Answer 2

• contains tables foreign keys?

• don't use select *, but join table

maybe:

SELECT serial, table1.data1,table2.data2,table3.data3 FROM list
LEFT JOIN table1 USING serial
LEFT JOIN table2 USING serial
LEFT JOIN table3 USING serial
GROUP BY list.serial

• create keys

maybe process queries separately and link data in PHP