Neo4j--- performance optimization

One 、 Basic tuning

1、cql Statement performance troubleshooting

（1）EXPLAIN + Executive cql sentence
　　　　 Just view the execution plan , Do not execute statements

（2）PROFILE + Executive cql sentence
　　　　 Run statement , And supervise the use of resources in the whole process

（3） Check index creation
　　　　:schema

2、 Use advice
　　1） Create indexes for related primary keys Grammar format CREATE INDEX ON: label ( Fields to be checked );

① Schema index ： Schema indexes are very similar to indexes in relational databases , Each index corresponds to a label and a set of attributes , Whether updating or deleting nodes , The index will be automatically updated or deleted , Therefore, this method of index creation is more applicable .
　　 　 Be careful ： Index status is ONLINE To take effect , Otherwise, the newly created index is invalid .

2） Create unique index , Avoid data warehousing checks , Cause performance waste （ You can verify the integrity and uniqueness of data before warehousing ）CREATE CONSTRAINT ON (n:ResNode) ASSERT n.id IS UNIQUE;

3）create and merge The difference between MERGE The order is CREATE Command and MATCH The combination of orders . MERGE = CREATE + MATCH Therefore, new nodes should be avoided merge;

4） Unnecessary data , Filter out as early as possible , Reduce the amount of post-processing data ;
5） Avoid returning entire nodes , Return the required data .

Two 、 Optimize neo4j The configuration file

Yes neo4j.conf File and modify accordingly . By adding jvm Virtual environment can improve the query speed of database , Cancel now neo4j About... In the configuration file dbms.memory.heap.initial_size=512m;dbms.memory.heap.max_size=512m Two lines of comments , And make appropriate modifications （ The larger the maximum heap memory, the better , But less than the physical memory of the machine ）.

3、 ... and 、 Advanced solutions solve performance problems

1、 Reasonably introduce middleware

eg： Involving content keyword search , have access to elasticSearch, Avoid going directly from neo4j Search for , If you can combine redis And memory cache to get detailed resource information , Search only returns the primary key , It can further improve query submission efficiency , The user experience has greatly improved ;
　　　　 Involving a large number of queries, you can use redis As neo4j A cache library for , It can improve performance , Can also avoid neo4j Resource consumption ;

2、 Sub database storage , make the best of neo4j Graph storage and query features ,neo4j Only graph storage and query functions are provided in , The attribute information of the node is saved in mongodb（ Or others NoSql In the database — Fields can be stored directly ） Make association query , Each in his element , Make full use of advantages , Optimize performance ;

3、 Use neo4j Enterprise high availability 、 Cluster solution .