Optimzing your queries can help them run more efficiently, which can save a significant amount of time. This article covers index optimization and index usage. It is excerpted from chapter 13 of the MySQL Certification Guide, written by Paul Dubois et al. (Sams, 2005; ISBN: 0672328127).
An index helps MySQL perform retrievals more quickly than if no index is used. But indexes can be used with varying degrees of success, so you should keep several index-related considerations in mind when designing tables:
Declare an indexed column NOT NULL if possible. Although NULL values can be indexed, NULL is a special value that requires additional decisions when performing comparisons on key values. An index without NULL can be processed more simply and thus faster.
Avoid overindexing; don't index a column just because you can. If you never refer to a column in comparisons (such as in WHERE, ORDER BY, or GROUP BY clauses), there's no need to index it. Another reason to avoid unnecessary indexing is that every index you create slows down table updates. If you insert or delete a row, an entry must be added to or removed from each of the table's indexes. If you update a row, any change to indexed columns require the appropriate indexes to be updated as well.
One strategy the MySQL optimizer uses is that if it appears an index will return a large percentage of the records in the table, it will be just as fast to scan the table as to incur the overhead required to process the index. As a consequence, an index on a column that has very few distinct values is unlikely to do much good. Suppose that a column is declared as ENUM('Y','N') and the values are roughly evenly distributed such that a search for either value returns about half of the records. In this case, an index on the column is unlikely to result in faster queries.
Choose unique and nonunique indexes appropriately. The choice might be influenced by the type of a column. If the column is declared as an ENUM, there is a fixed number of distinct column values that can be stored in it. This number is equal to the number of enumeration elements, plus one for the '' (empty string) element that is used when you attempt to store an illegal value. Should you choose to index an ENUM column, you likely should create a nonunique index. A PRIMARY KEY allows only as many rows as the number of distinct enumeration values. A UNIQUE index enforces a similar restriction, except that unless the column is declared NOT NULL, the index allows NULL values.
Index a column prefix rather than the entire column. MySQL caches index blocks in memory to avoid whenever possible reading them from disk repeatedly. Shortening the length of indexed values can improve performance by reducing the amount of disk I/O needed to read the index and by increasing the number of key values that fit into the key cache. This technique is discussed further in section 22.214.171.124, "Indexing Column Prefixes."
Avoid creating multiple indexes that overlap (have the same initial columns). This is wasteful because MySQL can use a multiple-column index even when a query uses just the initial columns for lookups. For more information, see section 126.96.36.199, "Leftmost Index Prefixes."