MySQL Query Optimization (Takes way too long) - php

Question

I use the following prepared statement:
SELECT *
FROM
c_members,c_users,c_positions,c_done_meetings
WHERE
c_positions.POS_ID=c_users.POS_ID
AND c_members.CLUB_ID = ?
AND USER_POINTS >= ?
AND USER_POINTS <= ?
AND c_users.POS_ID LIKE ?
AND MEM_ACADEMY LIKE ?
AND MEM_SEX LIKE ?
AND MEM_GRADELVL LIKE ?
AND MEM_GPA >= ?
AND MEM_GPA <= ?
AND MEM_ARCHIVE = 0
GROUP BY
c_members.MEM_ID, c_members.CLUB_ID
HAVING
SUM(c_done_meetings.MEDONE_ATTEND = 'u') >= 1
ORDER BY
USER_POINTS DESC
However this query takes 21.971405982971 seconds to load 111 records. When I remove "Having SUM(...)" clause, the performance is 100% better. Is there a way I could optimize it better?
Edit: (Table structures)
CREATE TABLE IF NOT EXISTS `c_done_meetings` (
`MEM_ID` int(11) NOT NULL,
`CLUB_ID` int(11) NOT NULL,
`MEETING_ID` int(11) NOT NULL,
`MEDONE_ATTEND` varchar(1) NOT NULL COMMENT 'E=excused, U=unexcused, P=present',
UNIQUE KEY `unique` (`MEM_ID`,`CLUB_ID`,`MEETING_ID`),
KEY `MEETING_ID` (`MEETING_ID`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE IF NOT EXISTS `c_members` (
`MEM_ID` int(11) NOT NULL,
`CLUB_ID` int(11) NOT NULL,
`MEM_FIRST` varchar(50) NOT NULL,
`MEM_MIDDLE` varchar(50) DEFAULT NULL,
`MEM_LAST` varchar(50) NOT NULL,
`MEM_SEX` tinyint(1) NOT NULL COMMENT '0-Male 1-Female',
`MEM_EMAIL` varchar(100) DEFAULT NULL,
`MEM_GRADELVL` int(11) NOT NULL,
`MEM_ACADEMY` varchar(50) DEFAULT '',
`MEM_GPA` double DEFAULT '0',
`MEM_ADDRESS` varchar(500) DEFAULT NULL,
`MEM_CITY` varchar(100) DEFAULT NULL,
`MEM_STATE` varchar(100) DEFAULT NULL,
`MEM_ZIP` int(11) DEFAULT NULL,
`MEM_TELEPHONE` varchar(25) DEFAULT NULL,
`MEM_AP` tinyint(1) NOT NULL,
`MEM_HONORS` tinyint(1) NOT NULL,
`MEM_ESOL` tinyint(1) NOT NULL,
`MEM_HISP` tinyint(1) NOT NULL,
`MEM_WHITE` tinyint(1) NOT NULL,
`MEM_MULTI` tinyint(1) NOT NULL,
`MEM_NATIVE` tinyint(1) NOT NULL,
`MEM_BLACK` tinyint(1) NOT NULL,
`MEM_ASIAN` tinyint(1) NOT NULL,
`MEM_EXTRA` varchar(10000) DEFAULT NULL,
`MEM_ARCHIVE` tinyint(1) NOT NULL DEFAULT '0',
UNIQUE KEY `MEM_ID` (`MEM_ID`,`CLUB_ID`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE IF NOT EXISTS `c_positions` (
`POS_ID` int(11) NOT NULL AUTO_INCREMENT,
`CLUB_ID` int(11) NOT NULL,
`POS_NAME` varchar(20) NOT NULL,
`POS_DESC` varchar(500) NOT NULL,
`POS_ADMIN` tinyint(1) NOT NULL,
`POS_ATN_VIEW` tinyint(1) NOT NULL,
`POS_ATN_CHKIN` tinyint(1) NOT NULL,
`POS_ATN_FINALIZE` tinyint(1) NOT NULL,
`POS_MEM_VIEW` tinyint(1) NOT NULL,
`POS_MEM_ADD` tinyint(1) NOT NULL,
`POS_MEM_EDIT` tinyint(1) NOT NULL,
`POS_POS_VIEW` tinyint(1) NOT NULL,
`POS_POS_ADD` tinyint(1) NOT NULL,
`POS_POS_EDIT` tinyint(1) NOT NULL,
`POS_MEET_VIEW` tinyint(1) NOT NULL,
`POS_MEET_ADD` tinyint(1) NOT NULL,
`POS_MEET_EDIT` tinyint(1) NOT NULL,
`POS_EVENT_VIEW` tinyint(1) NOT NULL,
`POS_EVENT_ADD` tinyint(1) NOT NULL,
`POS_EVENT_EDIT` tinyint(1) NOT NULL,
`POS_EVENT_UPDATE` tinyint(1) NOT NULL,
`POS_REPORT_VIEW` tinyint(1) NOT NULL,
`POS_ARCHIVE_VIEW` tinyint(1) NOT NULL,
`POS_ANNOUNCEMENTS` tinyint(1) NOT NULL,
`POS_WEB_CUSTOM` tinyint(1) NOT NULL,
PRIMARY KEY (`POS_ID`),
UNIQUE KEY `UNIQUE_NAME` (`CLUB_ID`,`POS_NAME`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1 AUTO_INCREMENT=14 ;
CREATE TABLE IF NOT EXISTS `c_users` (
`MEM_ID` int(11) NOT NULL,
`CLUB_ID` int(11) NOT NULL,
`USER_PIN` int(11) NOT NULL,
`USER_POINTS` double NOT NULL,
`POS_ID` int(11) NOT NULL,
`USER_ARCHIVE` tinyint(1) NOT NULL DEFAULT '0',
UNIQUE KEY `MEM_ID` (`MEM_ID`,`CLUB_ID`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
Edit 2:
Yes all ids are indexed, the SUM(...) >= # calculates the number of missed meetings. # is a parameter set by the user (I just hard coded a 1 for testing)

Answer 1

You need to have indexes an all fields used in the WHERE clause, on all fields on which the tables are joined (you need to explicitly state your join conditions as right now, you are getting a Cartesian join), on all fields used for grouping and all fields used for sorting.
The last problem is the HAVING clause there. You are not going to be able to use an index at all for that since it is a calculated value. If this is a query you will use often in the system (i.e. not just for reporting), you might consider adding a field you can use as a flag for this filtering purpose. Whenever you set c_done_meetings.MEDONE_ATTEND = 'u' in any of your queries, you could also set this flag for the member or user or whatever this is associated with so that you have an easy field to filter on in a WHERE clause.
Outside of that, you might actually gain better performance by getting a reduced list of users or members with that value of u in a subselect and then join using that subselect as a table.
EDIT:
After seeing your actual table structure, I can clearly see where you need to add indexes. I am also wondering why you have tables c_users and c_members with the same exact primary key. Why would these not just be a single table?

Answer 2

Some things that pop out at me:
You use like quite a lot. Try to do something with your php code so that this isn't necessary. For example, mem_sex has a limited number of choices. Make the front end a radio button or dropdown so you send a value where you can use = instead of like.
Two, if you add a sum() to the select clause, and the appropriate group by clause, it should run faster. It's worth a shot.

Answer 3

You can denormalise you tables and add field to c_members that represent your
SUM(c_done_meetings.MEDONE_ATTEND = 'u') >= 1
But you need to update that field always, when updating c_done_meetings (can be done with trigger)
Also try to avoid LIKE conditions. Use = insead (it is possible at least for SEX)