How to avoid temporary table on group by with join? - mysql

Question

I'm having two tables say(for example), Department and Members
Department table description:
CREATE TABLE `Department` (
`code` int(10) DEFAULT NULL,
`name` char(100) DEFAULT NULL,
KEY `code_index` (`code`),
KEY `name_index` (`name`)
)
Department table values:
+------+-------------+
| code | name |
+------+-------------+
| 1 | Production |
| 2 | Development |
| 3 | Management |
+------+-------------+
Members table description:
CREATE TABLE `Members` (
`department_code` int(10) DEFAULT NULL,
`name` char(100) DEFAULT NULL,
KEY `department_code_index` (`department_code`),
KEY `name_index` (`name`)
)
Members table values:
+-----------------+----------------+
| department_code | name |
+-----------------+----------------+
| 1 | Ross Geller |
| 1 | Monica Geller |
| 1 | Phoebe Buffay |
| 1 | Rachel Green |
| 1 | Chandler Bing |
| 1 | Joey Tribianni |
| 2 | Janice |
| 2 | Gunther |
| 2 | Cathy |
| 2 | Emily |
| 2 | Fun Bobby |
| 2 | Heckles |
| 3 | Paolo |
| 3 | Mike Hannigan |
| 3 | Carol |
| 3 | Susan |
| 3 | Richard |
| 3 | Tag |
+-----------------+----------------+
I want to get the all the department code and name for the given set of users. As i just want the department names alone, I used the below query.
mysql> select Department.code, Department.name, Members.department_code from Department left join Members on (Department.code=Members.department_code) where Members.name in ('Rachel Green', 'Gunther', 'Paolo') group by Department.code;
+------+-------------+-----------------+
| code | name | department_code |
+------+-------------+-----------------+
| 1 | Production | 1 |
| 2 | Development | 2 |
| 3 | Management | 3 |
+------+-------------+-----------------+
This works fine and the "explain" gives me below execution plan.
+----+-------------+------------+------------+------+----------------------------------+-----------------------+---------+----------------------+------+----------+---------------------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+------+----------------------------------+-----------------------+---------+----------------------+------+----------+---------------------------------+
| 1 | SIMPLE | Department | NULL | ALL | code_index | NULL | NULL | NULL | 3 | 100.00 | Using temporary; Using filesort |
| 1 | SIMPLE | Members | NULL | ref | department_code_index,name_index | department_code_index | 5 | test.Department.code | 1 | 16.67 | Using where |
+----+-------------+------------+------------+------+----------------------------------+-----------------------+---------+----------------------+------+----------+---------------------------------+
But the "group by" uses temporary table which may degrade the performance if the Members table contains a lot of rows. Though I guess some ideal indexing would help out here, i can't get the proper idea. Any help will be appreciated.
Thanks in advance!

Answer 1

You can avoid the group by over all the data using a subquery:
select d.code, d.name, d.department_code
from Department d
where exists (select 1
from Members m
where d.code = m.department_code and
m.name in ('Rachel Green', 'Gunther', 'Paolo')
);
With an index on members(department_code, name), this should be much faster.