Yes, job tracker and task tracker are present in different machines. The reason is job tracker is a single point of failure for the Hadoop MapReduce service. If it goes down, all running jobs are halted.