SQL查询联接表和having子句

Question

I am writing a query in SQL to pull a player's home run totals in their age 35 season who have hit over 500 home runs in their career.

SELECT b.playerID, b.yearID, b.HR
FROM batting b
JOIN master m ON m.playerID = b.playerID
WHERE b.yearID - m.birthYear = '35'
HAVING SUM(b.HR) > 500

This query times out while executing. I have successfully created a query to return a player's home run total in a specific age season. I have also successfully created a query to return players in the 500 home run club.

When I try to combine them something makes it time out and I cannot determine why.

Here is a query that works well:

SELECT b.playerID, b.yearID, b.HR
FROM batting b
JOIN MASTER M ON b.playerID = m.playerID
WHERE b.yearID - m.birthYear = 35 AND b.yearID = 2015
ORDER BY b.HR DESC

Now if I could only incorporate returning only those players who have hit 500 career home runs in this result. Only 500 home run hitters HR total in 2015.

Answer 1

The most likely explanation is that the execution plan chosen by the optimizer isn't as efficient as it could be.

What we're not seeing is what indexes are available on these tables.

One thing that stands out about the query is this:

WHERE b.yearID - m.birthYear = '35'

MySQL is going to take every row from master with a given player_id and match that to every row from batting with that same player_id (because of the equality join predicate)

  ON m.playerID = b.playerID

And then MySQL has to take that set of combined rows and then calculate this expression

    b.yearID - m.birthYear

And then take the result from that and compare it to '35'.

Assuming that the playerID column is unique on the master table

We'd prefer to see the query predicates written in a form that can take advantage of an index on batting that has leading columns of (playerID,yearID) .

 SELECT b.playerid
      , b.yearid
      , SUM(b.hr) AS hr
   FROM master m
   JOIN batting b
     ON b.playerid = m.playerid
    AND b.yearid   = m.birthyear + 35
  GROUP BY b.playerid, b.yearid
HAVING SUM(b.hr) > 500
 ORDER BY SUM(b.hr) DESC

To get rows returned for each player, you are going to need a GROUP BY clause. And to get the total home runs, you are going to need a SUM() aggregate in the SELECT list.

For optimal performance of the query, you'd want a covering index

... ON batting (playerid, yearid, hr)

If playerid is not unique on master table, then the query isn't going to guarantee the value you expect for SUM(b.hr), the value could be double, triple, etc. of what is expected.

Use EXPLAIN to see the execution plan.

Also beware of implicit datatype conversions that can have a negative impact on an execution plan. We're assuming that the datatype of the playerid column in both tables matches, and that the datatype of the yearid and birthyear columns is numeric.

EDIT

My original answer was focused on the reasons your query "timed out", and I missed the specification for the result you want to achieve:

Return players that have career HR total over 500, and return the year total HR for a specific year for each of those players.

(I'll set aside a discussion of appropriately determining the "year" in which a player turns 35 years of age, and use the criteria from the original query.)

One approach is to use conditional aggregation . Use an expression that returns HR when a condition is TRUE, and otherwise return 0 or NULL. And then wrap that expression in a SUM aggregate in the SELECT list.

If we want to return only rows for players with career HR total over 500 and that have at least one row in batting for the specified year...

 SELECT b.playerid
      , MAX(IF(b.yearid = m.birthyear + 35,b.yearid,NULL)) AS yearid
      , SUM(IF(b.yearid = m.birthyear + 35, b.hr, 0)) AS year_hr
   FROM master m
   JOIN batting b
     ON b.playerid = m.playerid
  GROUP BY b.playerid
HAVING SUM(b.hr) > 500
   AND MAX(IF(b.yearid = m.birthyear + 35,b.yearid,NULL)) IS NOT NULL
 ORDER BY ... 

To return rows for every player that has a career HR total over 500, even when there are no rows in batting for the specified yearid , we can tweak the query to omit the second condition in the HAVING clause, and use expression m.birthyear + 35 in the the SELECT list

 SELECT b.playerid
      , MAX(m.birthyear + 35) AS yearid
      , SUM(IF(b.yearid = m.birthyear + 35, b.hr, 0)) AS year_hr
   FROM master m
   JOIN batting b
     ON b.playerid = m.playerid
  GROUP BY b.playerid
HAVING SUM(b.hr) > 500
 ORDER BY ... 

Note that players that have a career HR total of exactly 500 will be excluded.