NodeManager

Resource Manager

Data Node

NameNode

True

False

RequirePythonversion no lower than2.6.6,not higher than2.7.0

Client must havesetuptools, and the version is not lower than5.0

hive_python_clientscript provides direct executionSQLfunctions, such ashive_python_client ' show tables '

PythonSecondary development does not support secure clusters

Mass data storage

Lots of small file storage

large file storage

Streaming file storage

SupervisorAfter the process exits, it can beNimbusAutomatically identify and pull up without affecting running business.

WorkerCan be used after abnormal exitSupervisorPulls up automatically without manual intervention.

When a node fails, the tasks on the node will be reassigned to other normal nodes without manual intervention.

NimbusAfter the failure, the backupNimbusIt will take over automatically without affecting the running business.

stream processing

high performance

Scalable

High fault tolerance

enabledAckerLater,StreamingIt will identify the failure to sendTupleAnd automatically resend without human intervention.

AckerThe message that will identify the processing timeout or processing failure isfail.

fromspoutstarted, formedTupleFailure of any link in the tree will mark the entire tree as failed.

The application needs to be inspoutoffail() interface method to implement message resend logic.