If the concept of function-based indexing is new to you, you might be tempted to just index everything, but this is in fact the very last thing you should do. The reason is that every index causes ongoing maintenance. Function-based indexes are particularly troublesome because they make it very easy to create redundant indexes.
The case-insensitive search from
above could be implemented with the
SELECT first_name, last_name, phone_number FROM employees WHERE LOWER(last_name) = LOWER('winand')
A single index cannot support both methods of ignoring the case.
We could, of course, create a second index on
LOWER(last_name) for this query, but that would mean the
database has to maintain two indexes for each
delete statement (see also Chapter 8, “Modifying Data”). To make one index suffice, you
should consistently use the same function throughout your
Unify the access path so that one index can be used by several queries.
WarningSometimes ORM tools use UPPER and LOWER without the developer's knowledge. Hibernate, for example, injects an implicit LOWER for case-insensitive searches.
Always aim to index the original data as that is often the most useful information you can put into an index.