JobTracker & TaskTrackers in Hadoop

The primary function of the job tracker is resource management (managing the task trackers), tracking resource availability and task life cycle management (tracking its progress, fault tolerance etc.)

The task tracker has a simple function of following the orders of the job tracker and updating the job tracker with its progress status periodically.

The task tracker is pre configured with a number of slots indicating the number of tasks it can accept. When the job tracker tries to schedule a task, it looks for an empty slot in the task tracker running on the same server which hosts the data node where the data for that task resides. If not found, it looks for the machine in the same rack. There is no consideration of system load during this allocation.

HDFS is rack aware in the sense that the name node and the job tracker obtain a list of rack ids corresponding to each of the slave nodes (data nodes) and creates a mapping between the IP address and the rack id. HDFS uses this knowledge to replicate data across different racks so that data is not lost in the event of a complete rack power outage or switch failure.

Job Performance - Hadoop does speculative execution where if a machine is slow in the cluster and the map/reduce tasks running on this machine are holding on to the entire map/reduce phase, then it runs redundant jobs on other machines to process the same task, and whichever task gets completed first reports back to the job tracker and results from the same are carried forward into the next phase.

Fault Tolerance - The task tracker spawns different JVM processes to ensure that process failures do not bring down the task tracker.

The task tracker keeps sending heartbeat messages to the job tracker to say that it is alive and to keep it updated with the number of empty slots available for running more tasks.

From version 0.21 of Hadoop, the job tracker does some check pointing of its work in the file system. Whenever, it starts up it checks what was it upto till the last CP and resumes any incomplete jobs. Earlier, if the job tracker went down, all the active job information used to get lost.

The status and information about the job tracker and the task tracker are exposed vis jetty onto a web interface.

Olap on Hadoop