Two minutes, talk about some wrong understandings of MySQL index

Hello , I am a yes.

Today, let's talk about MySQL InnoDB Some easily misunderstood points of the index .

Don't talk much , Direct departure ！

You may not use the index you think you will use

In order that the story can develop smoothly , Let's start with a watch , DDL as follows ：

Now I want to execute this statement SELECT * FROM yes.t1 where f3 = 11;

Which index do you think is used ？

Many students may feel that although there is idx_f2_f3 This joint index , But the query criteria are f3, No, f2, Does not conform to the leftmost matching principle , Therefore, it is obvious that only full table scanning can be performed .

Let's come. explain once , have a look ：

You can see the result ,key yes idx_f2_f3 , So this query uses idx_f2_f3 This index ！

Someone here may not understand , How can joint indexing be used if it does not conform to the leftmost indexing principle ？

from explain As a result , Do use the union index , But it's type yes index, therefore It's not the kind of index you think , It is equal to the full scan of the joint index to obtain the results .

It's ok if you don't understand now , Listen to the following explanation .

What exactly is using an index ？

We all know InnoDB It is based on clustering index to build data , Without creating other secondary indexes , You must and can only use cluster index to query data , Because the data is only on this index , Then you don't need this index to find data. What do you use ？

If you build other secondary indexes , Then you may not need cluster index to find data , Instead, only the upper secondary index can directly get the desired data ( This is the example of the above query ).

in summary , you are here InnoDB Find any data , Can only use index .

Then why do some people say that they don't follow the index on weekdays , After walking, the whole table is scanned ？

We all know the default InnoDB The index structure of is B+ Trees , This tree has a feature , Its leaf nodes are connected by linked lists .

So as long as we find primary key The leftmost leaf node , Then traverse the scan in right order , You can get all the records of this form , This scanning method is called full table scanning .

Tips: ： I don't know MySQL Is the location of the leftmost leaf node recorded or the leftmost leaf node found according to the root node , Anyway, it means that , It does not affect our understanding of full table scanning .

Then what is index ？

Look at the red query route , It can use the fast search ability of the tree to locate the page where the data is located , Such a query can count as an index .

Then push back from the primary key index to the secondary index , It's the same thing . It can use the characteristics of secondary index tree to find data , Is to use the index .

But if you can't use the fast search of the secondary index tree , Is there such a thing as full table scanning , The operation of directly scanning all leaf nodes of the secondary index ？

yes , we have , This is our example above , I'll move down the screenshot of the execution result ：

This query MySQL Full scan secondary index idx_f2_f3 To directly return the required data .

The reason is that the secondary index already contains all the data required by the query :f1、f2、f3, So you can scan the secondary index directly .

You can see explain Of type=index, That is, full scan secondary index .

This is also called using the index .

At this time, we use f2 As a query condition , This conforms to the leftmost matching principle , Let's see explain What's the difference? ：

You can see that it is still used idx_f2_f3 , It's just type become ref , This ref It shows that you can quickly query the results when matching with constants .

This is also called using the index .

Okay , Let's sum up what it means to use an index ：

• Use the primary key index to quickly find
• Fast search with secondary index
• Full scan the secondary index to find

The above three kinds are what we usually call using indexes , Then the full scan of the primary key index is what we call the full table scan （ Equivalent to not using an index , But we know that it actually uses the primary key index, right ？）.

thus , I think you should already know , What exactly is using an index .

Little doubt

About MySQL Some understanding of the index is over , But look at the above example , Some students may be a little confused ：

Since they all need full scanning , Why scan the secondary index , Instead of the primary key index ？

Because the primary key index stores all the data , Including transactions related to data rows ID、 Rollback pointer, etc .

The secondary index does not contain these , It only has primary key value and index column value .

Therefore, the overhead of directly reading the secondary index will be smaller , therefore When the index can overwrite the return value , Generally, the secondary index will be selected to find .

For example, I add a field to the table above ：f4.

Now let's execute SELECT * FROM yes.t1 where f3 = 11;：

You can see , At this time, the secondary index is not used , Instead, the whole table is scanned directly .

The reason is that the secondary index cannot overwrite the return value , Because there are more f4 Field of .

If used idx_f2_f3 This index , Want to get f4 Value , You also need to find the primary key in the primary key index , Then we can get f4 Value , In this way, the efficiency of returning the table is low .

Of course, the result is not certain , Sometimes it is possible MySQL The statistics are wrong , Wrong index . because MySQL The execution plan is selected based on cost , Yes I/O Cost and CPU cost , Use whichever index has a low cost .

therefore , In the next article, I will analyze how the cost is calculated ！

Last

I believe you have fully understood what is the use of the index ？

Okay , That's all for today's article , Wait for my next article MySQL Cost analysis ！

You can click on the business card below , Pay attention to me ~

---

I am a yes, From a little to a billion, I'll see you next .