I only want one row from my tables. Is it better to do a JOIN or just make two select calls

Question

I got two tables I need one row containing all columns of table1 and only 1 column of table2. I am getting the row by using order by rand() limit 0,1;

I am wondering if it more efficient to select all columns of table1 then make another call to get that extra column from table 2. since i only need one record, i am worried it is doing extra work in the background to join the tables when I only need one row.

I am still developing locally and my computer is fast so I can't tell the difference, but when it goes live there will be many calls per second on paid servers so resource is precious

Answer 1

There is no efficiency in order by rand limit 0,1; it will not scale well. It is a design flaw out of the gate. If you insist on this then you would certainly not want to do a join since this will iterated over every row in the table. Minimally, you should pick a small sample of rows to pull a random one from:

$row=SELECT table_rows*rand() FROM INFORMATION_SCHEMA.TABLES WHERE foo and bar
SELECT * FROM foobar LIMIT $row, 1