Question

I’m making a dictionary webapp. The user will search for words. Would it be faster to do this?

SELECT * from definition WHERE word LIKE "house";

or…

SELECT * from definition WHERE word_hash LIKE md5("house");

In the second example, I store the md5() value of words in the word_hash field. Of course, “word” and “word_hash” are indexes.

Update: sometimes, the word field could be more than 1 word. Example: Sacré Bleu

Answer 1

Skipping LIKE completely would be faster. Added the lower case version of word as word_lc, index word_lc, and then do:

select * from definition where word_lc = lower(word_you_want)

Using LIKE without any % or _ wildcards is just a case insensitive equality test so you should go straight to a case insensitive comparison that can and will take advantage of an index. Also, as usual, say what you mean so the computer can do what you want it to do.