Large Scale Network-Centric Distributed Systems

21

On the Design of a Budget-Conscious Adaptive Scheduler for Handling Large-Scale Many-Task Workflow Applications in Clouds

Bharadwaj Veeravalli, Lingfang Zeng, and Xiaorong Li

Contents

21.1 Introduction

21.2 Related Work and Motivation

21.3 System Model and Problem Setting

21.3.1 System Model

21.3.2 Many-Task Workflow Scheduling Problem

21.4 Proposed Scheduling Algorithm

21.4.1 Static ScaleStar

21.4.1.1 Initial Assignment Phase

21.4.1.2 Task Reassignment Phase

21.4.1.3 DeSlack Policy

21.4.2 Dynamic Adaptive Strategy

21.5 Performance Evaluation and Results

21.5.1 Evaluation Methodology

21.5.2 Real-World MTW Applications

21.5.3 Synthetic MTWs

21.5.4 Static Strategy Evaluation

21.5.5 Dynamic Strategy Evaluation

21.6 Conclusion

References

21.1 Introduction

Clouds have emerged as a new service provisioning model and are capable of supporting diverse services (servers, storage, network, and applications) over a global network. For Cloud computing–based services, users consume the services when they need to and pay only for what they use. With an economic incentive, Cloud computing encourages individuals and organizations to remotely access resources offered by Cloud Service Providers (CSPs). Therefore, it facilitates individuals and organizations in developing their own core activities without maintaining and developing fundamental infrastructure. Cloud infrastructure enables users to consume services transparently over a secured, shared, scalable, sustainable, and standard worldwide ...