[teacher Zhao Yuqiang] index in mongodb (Part 1)

Index is the most effective way to improve query efficiency . Index is a special data structure , The index stores part of the data in a form that is easy to traverse （ Such as ： A specific field or set of field values ）, The index sorts the stored values according to certain rules , And the index is stored in memory , It's very fast to retrieve data from the index . If there is no index ,MongoDB Every document in the collection must be scanned , This kind of scanning is very inefficient , Especially when the amount of data is large .

One 、 The basics of indexing

Here's a relational database Oracle For example , Introduce the basic principle of index , As shown in the figure below ：

It can be seen from the above picture that , The essence of an index is actually a table of contents for a book . When querying data in a table , Query directory first （ Indexes ） Line address in , Then query the data in the table through the row address , To improve query performance .

The figure below illustrates the MongoDB in , How indexes work in queries and sorts ？

Through this example , You can clearly see that the index stores a specific field or a collection of several fields , And sort them according to certain rules . When you create a collection ,MongoDB Automatically in _id Create a unique index on , Because it's unique , So it can prevent repetition _id Values are inserted into the collection . adopt getIndexes You can look it up MongoDB Index information on a set , As shown in the figure below .

When there is no index , By looking at the execution plan , You can see the query process , as follows ： Inquire about ：10 Department No , Wages less than 3000 Documents .

So how to create a simple index ？ Attention from mongoDB 3.0 Start ensureIndex Abandoned , In the future, it's just db.collection.createIndex An alias for .

Now in deptno and sal Build an index on , And review the execution plan ：db.emp.createIndex({“deptno”:1,“sal”:-1})

Be careful ： Except that it can be used explain() Generate execution plan out of plan , There are also several optional parameters , as follows ：
.explain(“allPlansExecution”)
.explain(“executionStats”)
.explain(“queryPlanner”)

Two 、 Index type one ： Single key index （Single Field）

The single key index is the most common index , And _id Different indexes , Single key indexes are not automatically created .

 Prepare the data ：
db.testindex1.insert({"_id":1,"zipcode":1034,"location":{state:"NY",city:"New York"}})

 Create a single key index on a single column ：
db.testindex1.createIndex({"zipcode":1})

 Create a single key index on nested Columns 
db.testindex1.createIndex({"location:state":1})

 Create a single key index on an embedded document 
db.testindex1.createIndex({"location":-1})
 This will put location As a whole .

3、 ... and 、 Index type two ： Multi key index （Multikey Index）

A multi key index is created in the same way as a single key index , The difference is the value of the field . Value has multiple records , Such as arrays .

Pictured above , Create a multi key index based on an array on a collection , And the array is an embedded document .

 Prepare the data ：
db.testindex2.insertMany([
{ _id: 5, type: "food", item: "aaa", ratings: [ 5, 8, 9 ] },
{ _id: 6, type: "food", item: "bbb", ratings: [ 5, 9 ] },
{ _id: 7, type: "food", item: "ccc", ratings: [ 9, 5, 8 ] },
{ _id: 8, type: "food", item: "ddd", ratings: [ 9, 5 ] },
{ _id: 9, type: "food", item: "eee", ratings: [ 5, 9, 5 ] }])

 Based on ratings Column creates a multi key index :
db.testindex2.createIndex( { ratings: 1 } )

 Query array for 5,9 Documents 
db.testindex2.find( { ratings: [ 5, 9 ] } )

 Here's the execution plan 
db.testindex2.find( { ratings: [ 5, 9 ] } ).explain()