Spark Architecture
- Driver is a JVM process that hosts SparkContext for a Spark application
- Executor is a distributed agent that executes Spark tasks
- Master and Slaves / Workers
- Switch to Mastering Apache Spark 2
Dynamic Allocation of Executors
- Dynamic Allocation is a Spark feature that allows for adding or removing executors dynamically based on the workload
- Switch to Mastering Apache Spark 2
Dynamic Allocation Demo - run-example
- Spark on YARN with YARN-specific ExternalShuffleService
- run-example is spark-submit to run Spark examples, e.g. SparkPi
- web UI's Timeline
- Switch to Mastering Apache Spark 2
Dynamic Allocation Demo - spark-shell
- Programmable Dynamic Allocation in spark-shell
- Switch to Mastering Apache Spark 2