Spark divides the stages according to the dependencies of the RDD. The scheduler starts from the end of the DAG graph and traverses the entire dependency chain in reverse. It will disconnect when it encounters a narrow dependency, and add it to the current stage when it encounters a wide dependency.

Streaming of Fusioninsight HD is weakly dependent on Zookeeper, even if Zookeeper fails, Streaming can provide services normally.

What is the data scale unit of traditional data processing?

Which parts of the data need to be read to perform the HBase data reading service? (Multiple choice)

In Streaming, one accurate message reliability level is achieved through the ACK mechanism.