This blogs explains how you can improve the performance of your select queries by properly using indexes.
Consider the following query on a data base with no indexes.
select ci.Name, co.Name, cl.Language from city ci, country co, countrylanguage cl where ci.CountryCode=co.Code and cl.CountryCode=co.code and ci.CountryCode = 'USA';
Lets do an explain on the above query.
explain ci.Name, co.Name, cl.Language from city ci, country co, countrylanguage cl where ci.CountryCode=co.Code and cl.CountryCode=co.code and ci.CountryCode = 'USA'; Explain command gives out following result
Which means 1004056074 (243107183858) rows had to be examined
Lets add an index to CountryCode column of CountryLanguage table
create index idx_CountryCode ON CountryLanguage (CountryCode); Executing the explain command again gives following result.
Now if you see row scan for CountryLanguage table alone has reduced from 1071 to 12 rows only so in total row scan has been reduced to 14768424
Now lets add an index on Code column of Country table
create index idx_Code ON Country (Code); Executing the explain command now gives following result.
Now total row scan has been reduced from 14768424 to 46296
Now we will add index on CountryCode column of City table.
create index idx_CountryCode ON City (CountryCode); Following is the result of explain command.
Total row scan has been reduced to 3276 rows only which is a lot faster then scanning 1004056074 on non indexed columns.
Things to keep in mind
Indexs are very powerful but they should be used smartly, creating too many indexes can slow down insert, update and delete process. On which column to add index can be identified by joins in your query, but if a join is with a table which has just 3-4 rows adding index won't help much.