Flink learning 4:flink technology stack

Catalog ：

 

1.flink The technical architecture is layered

The core is divided into three layers ： Physical deployment layer ,Runtime layer ,API & liberare layer  

Physical deployment layer ：

1. flink Support single node deployment ,, Single jvm management

2. Cluster mode is also supported ,standalone Pattern ,YARN Pattern

3. It can also be deployed on the cloud , Google cloud, Amazon cloud, etc

Runtime layer

As long as it is the upper layer API Provide services , It's also flink Core code layer .

API & liberare layer  

API The module is mainly divided into 2 part ：

Stream processing module ：DataStream and Batch module ：DataSet

stay API The two modules are abstracted out , Different libraries .

CEP： Complex event processing library for stream processing

SQL&Table library ： You can handle streams , You can also process batches .

FlinkML： Batch based machine learning library .

Gelly： Graph calculation library based on batch processing .