JobTracker

• Manages MapReduce jobs by scheduling and monitoring Map and Reduce tasks.
• Handles task failures and retries failed tasks if necessary.
• Coordinates with TaskTrackers or NodeManagers to allocate resources for tasks.

TaskTracker

• Executes tasks assigned by the JobTracker on each node in the cluster.
• Monitors task health and progress and reports to the JobTracker.
• Replaced by NodeManager in YARN (MapReduce v2).

ResourceManager (YARN)

• Replaces JobTracker in MapReduce v2.
• Manages resources across the Hadoop cluster and allocates them to applications.
• Coordinates with NodeManagers for resource tracking and scheduling.

NodeManager (YARN)

• Manages resources of individual nodes and executes tasks.
• Reports resource usage (memory, CPU) and task health to the ResourceManager.

Mapper

• Processes input data and generates key-value pairs as intermediate results.
• Transforms each InputSplit into intermediate key-value pairs.

Shuffle and Sort

• Transfers intermediate data from Mappers to Reducers.
• Sorts and groups key-value pairs by key to prepare for reduction.

Reducer

• Processes sorted key-value pairs and generates the final output.
• Aggregates values associated with each key and performs operations like summing or filtering.