4.2.2. Algorithm of locality-aware scheduler
This study implements the locality-aware scheduler in JobTracker.
Before assigning tasks, JobTracker uses Eq. (1) to calculate the
weight of data interference on each node, which has empty slots to
execute the task. JobTracker then picks up a data node with a
smallest weight of data interference and dispatches the task to the
TaskTracker queue within the data node. Algorithm 3 shows the details
of the locality-aware scheduling algorithm. The scheduler not
only calculates the weight of data interference to avoid rare resource
allocated in an easy way but also introduces the concept of weight of
data interference to enhance the data locality in the MapReduce
framework. The locality-aware scheduler thus reduces the overhead
of data transmission, which is caused by the network delay, and the
performance of cloud platform also is improved by the localityaware
scheduler.
This