Confusion on creating index to improve join performance in mysql - mysql

Question

I read many posts on forum, but still I have confusion on creating index to speed up join queries in mysql, here is my doubt
I have two tables, one is category table which just contains few thousand lines, and contains all information about data, and another one is geo_data table which contains huge amount of data, I join geo_data table based on 2 keys s_key1 and s_key2. Following is structure of table
category table
CREATE TABLE `category` (
`Id` int(11) NOT NULL AUTO_INCREMENT,
`s_key1` int(11) DEFAULT NULL,
`s_key2` int(11) DEFAULT NULL,
`STD_DATE` datetime DEFAULT NULL,
`LATITUDE` float DEFAULT NULL,
`LONGITUDE` float DEFAULT NULL,
`COUNTRY_CD` varchar(15) DEFAULT NULL,
`INSTR_CODE` varchar(15) DEFAULT NULL,
`CANADACR_CD` varchar(15) DEFAULT NULL,
`PROBST_T` varchar(15) DEFAULT NULL,
`TYPE` varchar(15) DEFAULT NULL,
PRIMARY KEY (`Id`)
) ENGINE=MyISAM AUTO_INCREMENT=32350 DEFAULT CHARSET=latin1;
geo_data table
CREATE TABLE `geo_data` (
`s_key1` int(11) DEFAULT NULL,
`s_key2` int(11) DEFAULT NULL,
`MAGNETIC` float DEFAULT NULL,
`GRAVITY` float DEFAULT NULL,
`BATHY` float DEFAULT NULL,
`CORE` float DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=latin1;
I have many tables like geo_data table which contains s_key1, s_key2 and other columns, in my application I often use fields std_date,latitude,longitude,country_cd,type from category table
I do inner join, sometimes left join depending on the requirement, for example my query looks like below
SELECT
c.s_key1,
c.s_key2,
c.std_date,
c.latitude,
c.longitude,
g.magnetic,
g.bathy
FROM
category c, geo_data g
WHERE
c.s_key1 = g.s_key1 && c.s_key2 = g.s_key2;
and sometimes my where clause will have something like this too
WHERE
c.latitude between -30 to 30 AND
c.longitude between 10 to 140 AND
c.country_cd = 'INDIA' AND
c.type = 'NON_PROFIT';
So what's the right way of creating index to speed up my query, whether below one right ? please someone help
create index `myindex` on
`category` (s_key1,s_key2,std_date,latitude,longitude,country_cd)
create index `myindex` on
`geo_data` (s_key1,s_key2)
and One more doubt whether both tables (category,geo_data) should have index key to speed up performance or only geo_data table ?

Answer 1

From the where condition it makes sense to simplify the first index as:
create index `myindex` on
`category` (s_key1,s_key2)
The original however can improve the performance in terms that it doesn't have to access the full table row to get the other values. However it makes the index bigger and therefore slower. So it depends on whether this is optimization for only this query or there are more of them which use only the s_key1 and s_key2 (or with combination with other columns).
Regarding the clarification - for lat/lng check it will make sense to move std_date after lat/lng (or remove completely):
create index `myindex` on
`category` (s_key1,s_key2,latitude,longitude,std_date,country_cd)