Neo4j learning notes

Ne04j

brief introduction

• A kind of Native graph database, namely Native graph database system . Use the customized data structure to store the map on the local hard disk , Instead of being stored in database relational tables
• Open source Java Realization

Neo4j Storage

• The vertices ：node
• edge ：relationship
• Vertices and edges can have multiple key-value Yes ： Become property

Neo4j Use

• Cypher：Declarative query language（ Declarative query language ）
• Traversal：Embedded Java lib

Cypher

Node

(name:type, {
    property_key:value, property_key:value...})

Ahead name:type in ,
name It means that we should Node The name of
type That is Node The type of

In the following brace , That is, multiple key-value Composed of property

Example ：

//  Only name
(Jerry)
//  Yes name, And there are type
(Jerry:Dog)
//  Yes name, And Tang Dynasty type, also property
(Jerry:Dog, {
    name:"Jerry", sex:"male", age:"1"})
( emma :Vehicle,{
    name:" emma "})

Relationship

-[name:type, {
    property_key:value, property_key:value...}]->

And Node The main difference is that the outermost parentheses are not parentheses , It's brackets , There are combined characters like arrows outside the brackets .

-[:Cycling]->
-[:Cycling,{
    place:home, time: morning}]->

File store

Store all... Separately Node、Relationship and Property The file of

• Node store、Ralationship store、Property store file
• unique id： Every node,relationship,property All have their own unique id

Ralationship Double linked list

• The same node Of relationship Form a two-way linked list
• The link pointer is relationship id
• Corresponding node Record the first in the linked list relation id

Property One way linked list

• The same node/relationship Of property Form a one-way list
• The link pointer is next property id
• The first in the linked list property id There is a corresponding node/relationship in

Memory buffer

OS Your file system has a buffer pool , As shown in the table on the page

Node/Rel cache

• Neo4j Yes node,relationship and property buffer
• The internal data structure is based on node and relationship For the main unit , and Property It is added to node and relationship On

Writing documents

• One transaction（ It's business , An operation request to the database ） Write and save first , until transaction.finish() when , To really try to modify the data
• Adopted transaction Log

HA, namely High Availability（ High availability ）

• Use multiple copies
• Master copy transaction log Send to the slave copy
• From copy replay log So as to perform the same operation