COROB: a controlled resource borrowing framework for overload handling in cluster-based service hosting center

Authors:
Yufeng Wang;Huaimin Wang;Dianxi Shi;Bixin Liu
Affiliations:
National University of Defense Technology, ChangSha, China;National University of Defense Technology, ChangSha, China;National University of Defense Technology, ChangSha, China;National University of Defense Technology, ChangSha, China
Venue:
OTM'05 Proceedings of the 2005 Confederated international conference on On the Move to Meaningful Internet Systems - Volume >Part I
Year:
2005

Citing 6
Cited 0

On the self-similar nature of Ethernet traffic (extended version)

IEEE/ACM Transactions on Networking (TON)
Performance Guarantees for Web Server End-Systems: A Control-Theoretical Approach

IEEE Transactions on Parallel and Distributed Systems
A capacity management service for resource pools

Proceedings of the 5th international workshop on Software and performance
OnCall: Defeating Spikes with a Free-Market Application Cluster

ICAC '04 Proceedings of the First International Conference on Autonomic Computing
Overload management as a fundamental service design primitive

EW 10 Proceedings of the 10th workshop on ACM SIGOPS European workshop
A SLA-Based resource donation mechanism for service hosting utility center

GCC'05 Proceedings of the 4th international conference on Grid and Cooperative Computing

Quantified Score

Hi-index	0.01

Visualization

Abstract

The paper proposes a resource framework COROB for overload handling in component application hosting center through dynamic resource borrowing among hosted applications. The main idea is to utilize the fine-grained idle server resource of other applications to partake of surging workload, while keeping the resource borrowing under control for not violating the SLA of the donor application. The contribution of the paper is two-fold: (1) a queuing analysis-based resource borrowing algorithm is proposed for overloaded applications to acquire as exact amount of resource as possible; (2) an adaptive threshold-driven algorithm is presented to drive the overload handling with threshold values adaptively tuned according to changing workload. Empirical data is presented to demonstrate the efficacy of COROB for overload handling and response time guarantee in a prototype service hosting cluster environment.