Deadline Based Map-reduce Workload Management in Multijob

Abstract: A scheduling algorithm and technique for managing multi-job Map Reduce workloads that relies on the ability to dynamically build performance models of the executing workloads, and uses these models to provide dynamic performance management using Adaptive scheduler. One of the design goals of the Map-Reduce framework is mainly based Adaptive scheduler to maximize data locality across working sets, in an attempt to reduce network bottlenecks and increase overall system throughput. Data locality is achieved when data is stored and processed on the same physical nodes. Sometime the server based executing workloads are not delivered to those particulars. Because, the multi-job network areas occurred some problem. So, the server storage is too high. In this paper, overcome this problem by the use of another server that is related to the main server. The problem of main server workload data executing to related server. Finally, the unreachable storage data delivered from related server to the particular receiver. So, every time free storage space and speed process in this server and also improve the server response time.

Keywords: Map Reduce, performance management, task scheduling