Left Join with Null Script Efficiency Explanation Needed - mysql

Question

Why would I use a LEFT JOIN in SQL in a FROM clause and attach a WHERE clause where the entity "is null"? I was told this is a very efficient script and I should learn the methodology behind it.
For example:
FROM
something
LEFT JOIN aRow a AND bRow b AND cRow c AND dRow d
WHERE
bRow.b IS NULL;

Answer 1

This kind of construct is used when you specifically want to know something like "a list of all customers who have never ordered anything" :
SELECT
customer.*
FROM
customers
LEFT JOIN
orders
ON
orders.customerid = customers.id
WHERE
orders.id IS NULL
Or to quote an old manager of mine: "Can you get the database to give me a list of everything that isn't in the database?"
Me> "Sure, can you give me a list of what things the database should tell you it doesn't have?"
Him> "How am I supposed to know that?"

Answer 2

This really is a fairly generic, non-RDBMS-specific question. The logic will apply to pretty much any flavor of SQL. And this is a technique that anyone who works with data queries should be familiar with.
For all intents and purposes (and moving past the flawed syntax in the OP), this is the same query as:
SELECT *
FROM table1
WHERE table1.col1 NOT IN (
SELECT table2.col1 FROM table2 WHERE table2.col2 = <filterHere>
)
When you are dealing with a couple of hundred rows, you may not see a significant difference in performance. But when you're working with just a few million rows in both tables, you will most definitely see a significant performance increase in
SELECT table1.*
FROM table1
LEFT OUTER JOIN table2 ON table1.col1 = table2.col1
AND table2.col2 = 42
WHERE table2.id IS NULL
Let's illustrate what is happening with these queries.
Create test tables.
CREATE TABLE table1 (col1 int, col2 varchar(10)) ;
INSERT INTO table1 ( col1, col2 )
VALUES (1,'a')
, (2,'b')
, (3,'c')
, (4,'d')
CREATE TABLE table2 (col1 int, col2 varchar(10)) ;
INSERT INTO table2 ( col1, col2 )
VALUES (1,'a')
, (3,'c')
This gives us
table1
col1 col2
1 a
2 b
3 c
4 d
table2
col1 col2
1 a
3 c
Now we want the columns that are in table1 but not in table2.
SELECT t1.col1, t1.col2
FROM table1 t1
WHERE t1.col1 NOT IN (
SELECT t2.col1 FROM table2 t2
)
We can't SELECT anything from table2, because that table is just a sub-query and not part of the whole query. It's not available to us.
This breaks down to
SELECT t1.col1, t1.col2
FROM table1 t1
WHERE t1.col1 NOT IN ( 1,3 )
Which further breaks down to
SELECT t1.col1, t1.col2
FROM table1 t1
WHERE t1.col1 <> 1
OR t1.col1 <> 3
These queries give us
col1 col2
2 b
4 d
That's a subquery broken down into 2 different OR statements to filter our results.
So lets look at a JOIN. We want all of the records on the left side, and only include those on the right side that match. So
SELECT t1.col1 AS t1_col1, t1.col2 AS t1_col2, t2.col1 AS t2_col1, t2.col2 AS t2_col2
FROM table1 t1
LEFT OUTER JOIN table2 t2 ON t1.col1 = t2.col1
With a JOIN, both tables are available to our SELECT, so we can see which records in tablel2 match up to those in table1. The above gives us
t1_col1 t1_col2 t2_col1 t2_col2
1 a 1 a
2 b NULL NULL
3 c 3 c
4 d NULL NULL
With the extra data, we can see that col1 for 2 and 4 don't match in the two tables. We can now filter those out with a simple WHERE statement.
SELECT t1.col1, t1.col2
FROM table1 t1
LEFT OUTER JOIN table2 t2 ON t1.col1 = t2.col1
WHERE t2.col1 IS NULL
Giving us
col1 col2
2 b
4 d
There's no subquery and just one statement in the filter. Plus, this allows the engine's optimizer to make a more efficient query plan.
It's impossible to see a difference in performance when we're only dealing with a couple of rows, but multiply these tables by a few million rows, and you will definitely see how much faster a JOIN can be.