Add string index and prefix index

Add a string index:alter table SUser add index index1(email);
Contains the entire string of each record in index index1

Add a prefix index:alter table SUser add index index2(email(6));
For each record in Index2, only the first 6 bytes

Use prefix index shortcomings and advantages

Disadvantages: The prefix index may increase the scan line, because the less n bytes in the prefix index, the less scan lines are small

advantage: That is to say, use the prefix index, define the length, you can save space, no additional increase in more
inquiry cost.

How to determine how long the prefix is ​​used

In fact, we pay attention to the establishment of an index.distinctionThe higher the distortion, the better. Because the higher the distortion, the less key value is repeated. Therefore, we can determine how many different values ​​in the statistical index can be judged to use.

Check for phrases:SELECT Count (Distinct String Field) AS L from Name;
segmentation The value of the prefix:Select Count (Distinct Left (Field Name, Front K)) AS column name from the FROM table;
Query the value of different prefixes in the DEV table

mysq1> select
->count(distinct left(cpuid，2as L2，
->count(distinct left(cpuid，3 as L3,
->count(distinct ieft(cpuid,4 as L4，
->count(distinct left(cpuid，6as L6,
->count(distinct left(cpuid，10) as L10,
->count(distinct left(cpuid，12))as L12
->from dev ;
+——-—-+-—--——+-—-——-+-—-———+------+-—--——+
|  L2 |  L3  |  L4  |  L6  |  L10  |  L2 |
+——-—-+-—--——+-—-——-+-—-———+------+-—--——+
| 750 | 1175 | 1191 | 1192 | 1195 | 1195 |
+-———-+-——--—+-——---+-———--+-—--—-+——-—--+
1 row in set (0.03sec)

As you can see from the results of the query $L \geq 4$L≥4The value corresponding to the time is basically close, so if you want to add a CPUID string field as an index, you only need to intercept the first 4 characters.