在SQL中，是不是查询越短，运行越快

Question

I'm very curious to know because when I run these two query in MySQL the shorter run faster

SELECT FirstName, LastName, City, State 
FROM Person
LEFT JOIN Address ON Person.PersonId = Address.PersonId;

and

SELECT FirstName, LastName, City, State 
FROM Person
LEFT OUTER JOIN Address ON Person.PersonId = Address.PersonId;

In addition, I want to ask if the same thing happen in the other RDBMS such as Posrgres, MS SQL Server, Oracle and SQLite?

Answer 1

LEFT OUTER JOIN is just a synonym for LEFT JOIN , and the execution plan for both of your queries should be identical. Therefore, I would attribute any difference in performance to things other than the use of OUTER in the left join syntax. You should check what other tasks might be running on your database during the test, and also what other processes might be running on the OS.

Answer 2

First, performance of a query is not based on how long it is. There are many "obvious" reasons for this. But perhaps the most is that the size of the data is determined by the size of tables in the FROM clause, and that has nothing to do with how many characters are in the name.

Second, LEFT JOIN and LEFT OUTER JOIN are synonyms for each other. No doubt, the OUTER takes a new compute cycles more on a modern computer to parse. But the difference would be measured in fractions of a microsecond (probably) and are pretty immeasurable.

Some of the things that affect the performance of queries are:

• The size of the data.
• Operations such as GROUP BY AND ORDER BY .
• JOIN s, particularly without indexes.
• And much, much more.

Answer 3

No, you cannot correlate the length of a SQL statement's text to its execution duration. It's more accurate to correlate a statement's execution duration to its execution plan with respect to the cardinality of the plan's row source operations.

Existence proof:

Take a billion row table, T, with a single column primary key, P

The full table scan query

select * from T

will take longer to execute than the unique index scan query (followed by a table access by index rowid):

select * from T where P = :pval

Except for some odd maintenance operations (eg global index rebuild over a partitioned table), the shorter query will no doubt take longer to execute than the longer query.