Efficient Scheduling Algorithm for Cluster Platforms

Title: Efficient Scheduling Algorithm for Cluster Platforms

Research Question: How can we develop an efficient scheduling algorithm that optimizes both makespan and weighted minimal average completion time for jobs submitted to a cluster platform?

Methodology: The researchers proposed a new scheduling algorithm called ID-IMAG, which is based on a batch policy with increasing batch sizes and smart selection of jobs in each batch. This algorithm was assessed through intensive simulation results and compared to a new lower bound obtained by relaxing an Integer Linear Programming (ILP) model.

Results: The ID-IMAG algorithm showed promising results in terms of both makespan and weighted minimal average completion time. The algorithm was implemented on an actual real-size cluster platform, demonstrating its practical applicability.

Implications: The ID-IMAG algorithm can be a valuable tool for managing and scheduling jobs on cluster platforms. It combines two complementary criteria - makespan and weighted minimal average completion time - to represent both user-oriented objectives and system administrator objectives. This can lead to improved performance and efficiency in job scheduling on cluster platforms.

Link to Article: https://arxiv.org/abs/0405006v1 Authors: arXiv ID: 0405006v1