Question

I have a table with about 1000 cities. Sometimes, I need to search by city name and have therefore decided to make it an index. What would be the best practice to decide on the index length? 6 seems like the optimal choice, because there would be almost no duplicates.

Example Table:

cityID  |  cityName         |  countyID
     1  |  Bethlehem        |     30
     2  |  Blairstown       |     38
     3  |  Bloomfield       |     32
     4  |  Bloomingdale     |     34
     5  |  Bloomsbury       |     30
     6  |  Bogota           |     31
     7  |  Boonton          |     33
     8  |  Botsford         |     44
     9  |  Bound Brook      |     35
    10  |  Branchburg       |     35
    11  |  Branchville      |     36
    12  |  Brantwood        |     32
    13  |  Briarcliff Manor |     25

cityID – Primary Key

countyID – Key

cityName (6) – Key

Am I doing this right?

When I tried running

EXPLAIN SELECT * 
FROM  'CITIES' 
WHERE  'cityName' =  'Branchburg'

I got the following results:

id: 1
select_type: SIMPLE
table: CITIES
type: ref
possible_keys: cityName
key: cityName
key_len: 20
ref: const
rows: 2
Extra: Using where

Shouldn’t “key_len” be 6?

Answer 1

It depends on character set. For example, if you are using utf8, it is like 6 * 3 = 18 … thought, I’m not sure where 2 bytes came from