Hadoop Map Reduce Job Scheduler Implementation and Analysis in Heterogeneous Environment
Journal Title: International Journal of Research in Computer and Communication Technology - Year 2015, Vol 4, Issue 3
Abstract
Hadoop MapReduce is one of the popular framework for BigData analytics. MapReduce cluster is shared among multiple users with heterogeneous workloads. When jobs are concurrently submitted to the cluster, resources are shared among them so system performance might be degrades. The issue here is that schedule the tasks and provide the fairness of resources to all jobs. Hadoop supports different schedulers than the default FIFO scheduler We started experiment on Hadoop FIFO, Fair and Capacity scheduler with heterogeneous workloads. Our aim is to compare the different job scheduler with heterogeneous workload and it is important to understand the task scheduler parameter, based on that we considered few parameter for the performance analysis.
Authors and Affiliations
Swathi Prabhu, Anisha P Rodrigues
Keywords
Related Articles
- EP ID EP28171
- DOI -
- Views 264
- Downloads 2
A Study on Security Advances for LEACH in WSN
Wireless Sensor Network consist of several autonomous sensor nodes that are capable of some sort of sensing power. Since the sensor nodes are placed in harsh environments, WSN is vulnerable to various types of attacks....
A Specialized Learning Framework for Aggregate Performance
Aggregate Performance is study of behaviour of a person who’s registered in a social network. More number of individuals are connection each other in this networks. The media is facing a major problem for finding the...
Optimum Image Filtering Algorithm Over The Unit Sphere
A statistical shape model, which includes a network of deformable curves on the unit sphere as reference framework for seeking geometric features such as high curvature regions and labels such features via a deforma...
Dynamic Approach for Secure Data Publishing in Mining
More than a few anonymization techniques, such as simplification and bucketization, have been deliberate for privacy protecting micro data publishing. Current work has shown that generalization loses substantial quant...
Highly Secured High Throughput Efficient VLSI Architecture for AES Implementations
The AES algorithm can be implemented in different styles at programming levels. The paper compares the hardware efficiency of different AES implementations with respect to their area, speed and power performance espec...