Running Hadoop in Different Modes
Standalone Mode
Configurations done on a single machine.Does not require major configuration changes.
No daemons are running.Everything runs in single JVM.
HDFS is not utilized in this mode, instead local file system is used for input and output.
Good for development and test with small data, but will not catch all errors.
Standalone mode is much faster than Pseudo-distributed mode.
Pseudo-Distributed Mode (Multiple JVM's in Single Node Cluster)
Single machine but cluster is simulated.
All daemons are running on one node itself in separate JVM.
Replication factory is one for HDFS.
Good for development and debugging.
Configuration is required for this mode
Single machine but cluster is simulated.
All daemons are running on one node itself in separate JVM.
Replication factory is one for HDFS.
Good for development and debugging.
Configuration is required for this mode
Fully Distributed Mode
All components are run in separate nodes.
Its Production environment.
Data are used and distributed across many nodes.
Good for staging and production.
All components are run in separate nodes.
Its Production environment.
Data are used and distributed across many nodes.
Good for staging and production.
No comments:
Post a Comment