Optimizing simple query

Question

This query will show you all of the unread topics in a given forum. The output of EXPLAIN EXTENDED was a bit alarming. I'm wondering if someone here can provide some insight into how I can optimize.

SELECT topic.*
FROM topic
INNER JOIN board ON topic.board_id = board.id OR topic.board_id = board.mirror_board_id
INNER JOIN category ON board.category_id = category.id
INNER JOIN group_assoc
 ON (
  group_assoc.board_id = board.id AND
  group_assoc.group_id IN (4,15,18,22) AND
  group_assoc.viewable = 1
 )
WHERE topic.last_post_time > 1288278402
AND category.forum_id = 2
AND board.id NOT IN(4,3)
AND NOT EXISTS (
    SELECT *
    FROM topic_read_assoc
    WHERE topic_id = topic.id
    AND member_id = 332
)
ORDER BY topic.last_post_time DESC

output:

id select_type table type possible_keys key key_len ref rows filtered Extra
1 PRIMARY category ref PRIMARY,forum_id_2 forum_id_2 4 const 5 100.00 Using temporary; Using filesort
1 PRIMARY board ref PRIMARY,mirror_board_id,category_id_2 category_id_2 4 source_forum.category.id 4 100.00 Using where
1 PRIMARY group_assoc ref board_id,board_id_2,board_id_3 board_id_3 4 source_forum.board.id 4 100.00 Using where; Using index
1 PRIMARY topic ALL board_id_2 NULL NULL NULL 2462 100.00 Range checked for each record (index map: 0x4)
2 DEPENDENT SUBQUERY topic_read_assoc ref topic_id topic_id 8 source_forum.topic.id,const 1 100.00 Using index

Answer 1

Create an index on topic (last_post_time) .

You can also remove LIMIT 1 from the EXISTS subquery, it's redundant.

Answer 2

Rather than use EXISTS for the group_assoc table, use an inner join on board for it (which is in effect happening anyway with "WHERE board_id = board.id". Put the filtering info. in the WHERE clause.

The same goes for topic_read_assoc - use an inner join on topic ON topic_id = topic.id.

Also, this is just to make it easier, you can use the IN syntax for Board.id so it's only one line like this:

WHERE ... board.id NOT IN (3,4)

EDIT> As Quassnoi rightly pointed out below, simply adding inner joins will cause duplicates. So use DISTINCT or GROUP BY what you want to see. The query should not be using SELECT * anyway by the way. If a column is added to that table, your code could be broken (not in the query itself, but by what is done with the results).

Answer 3

this would be my query for showing all unread topics in a forum

select * from forum_topic where forum_id = 1 and num_views = 0;

simple example follows:

-- TABLES

drop table if exists forum;
create table forum
(
forum_id int unsigned not null auto_increment primary key,
title varchar(255) unique not null,
num_topics int unsigned not null default 0
)engine=innodb;


drop table if exists forum_topic;
create table forum_topic
(
topic_id int unsigned not null auto_increment primary key,
forum_id int unsigned not null,
subject varchar(255) unique not null,
num_views int unsigned not null default 0,
num_replies int unsigned not null default 0,
key (forum_id)
)engine=innodb;

delimiter #

create trigger forum_topic_after_ins_trig after insert on forum_topic
for each row
begin
 update forum set num_topics=num_topics+1 where forum_id = new.forum_id;
end#

delimiter ;

-- STORED PROCEDURES

drop procedure if exists get_forum_topic;

delimiter #

create procedure get_forum_topic
(
in p_topic_id int unsigned
)
begin
    update forum_topic set num_views=num_views+1 where topic_id = p_topic_id;
    select * from forum_topic where topic_id = p_topic_id;
end #

delimiter ;

-- TEST DATA

insert into forum (title) values ('forum1'),('forum2');

insert into forum_topic (forum_id, subject) values 
(1,'forum 1 topic 1'), (1,'forum 1 topic 2'),
(2,'forum 2 topic 1'), (2,'forum 2 topic 2');

-- TESTING 

call get_forum_topic(1);
call get_forum_topic(3);
call get_forum_topic(2);
call get_forum_topic(3);
call get_forum_topic(2);
call get_forum_topic(3);

select * from forum;
select * from forum_topic;

select * from forum_topic where num_views = 0;