Cute pig--pig.

Pig provides an engine for parallel data flow processing based on Hadoop, which describes the data flow in pig Latin, which is equivalent to a directed acyclic graph (DAG), which represents the operator that handles the data, and the vectors between nodes represent the data flow.

Pig provides standard SQL-like operations that are easier to maintain than direct write MapReduce code, primarily for ETL, native data research, and iterative processing.

Pig what data to eat, but also on non-Hadoop parallel operation, easy to control and modify, performance faster.

The machine to which the user submits the Hadoop task can be extracted as a gateway machine.

The pig core is written in Java, so pay attention to the setting of the JAVA_HOME environment variable, the script that starts pig is bash, can either run locally or run on a Hadoop cluster, you need to know the location of the Namenode and Jobtracker of the cluster. You can also run on a cloud service (such as an EMR for AWS).

Grunt is the command-line interaction tool for pig, note that Pig_classpath's settings, in addition to the interactive input of Pig Latin, can also be used as a shell to access HDFs while providing command kill, exec, and run to control pig and MapReduce.

Cute pig--pig.