A large numeric value is stored in far fewer bytes than the equivalent string value. It is therefore faster to move and compare numeric data, so it's best to choose numeric columns for unique id's and other similar fields.

_{This page is licensed: CC BY-SA / Gnu FDL}

MEMORY tables are a good choice for data that needs to be accessed often, and is rarely updated. Being in memory, it's not suitable for critical data or for storage, but if data can be moved to memory for reading without needing to be regenerated often, if at all, it can provide a significant performance boost.

The MEMORY Storage Engine has a key feature in that it permits its indexes to be either B-tree or Hash. Choosing the best index type can lead to better performance. See Storage Engine index types for more on the characteristics of each index type.

_{This page is licensed: CC BY-SA / Gnu FDL}

Comparing String Columns

When values from different columns are compared, the comparison runs more quickly when the columns are of the same character set and collation. If they are different, the strings need to be converted while the query runs. So, where possible, declare string columns using the same character set and collation when you may need to compare them.

VARCHAR vs BLOB

ORDER BY and GROUP BY clauses can generate temporary tables in memory (see MEMORY Storage Engine) if the original table doesn't contain any BLOB fields. If a column is less than 8KB, you can make use of a Binary VARCHAR rather than a BLOB.

_{This page is licensed: CC BY-SA / Gnu FDL}