Cloud computing paradigm provides a range of attractive features such as resource elasticity, cost efficiency, and ease of management. Although much progress has been made in the dynamic resource management using control-based approaches in cloud computing environments, there are two significant research challenges. First, only single metric like mean response time (MRT) has been considered in the study of quality of service in individual applications. Second, to evaluate the performance of dynamic controllers, a single benchmark application is usually utilized in the previous work although the application may be driven by different synthetic workloads or utilization traces collected from production environment. This chapter analyses these two issues through experiments on a test bed with multi-tier applications hosted on multiple virtual machines. It evaluates the relationship between resource utilization and MRT for a three-tier application when the CPU shares of the virtual machines (VMs) are dynamically tuned through the utilization controllers.
[1]
Xiaoyun Zhu,et al.
Utilization and SLO-Based Control for Dynamic Sizing of Resource Partitions
,
2005,
DSOM.
[2]
Kang G. Shin,et al.
What does control theory bring to systems research?
,
2009,
OPSR.
[3]
Kang G. Shin,et al.
Adaptive control of virtualized resources in utility computing environments
,
2007,
EuroSys '07.
[4]
Chenyang Lu,et al.
Feedback utilization control in distributed real-time systems with end-to-end tasks
,
2005,
IEEE Transactions on Parallel and Distributed Systems.
[5]
F. Perros,et al.
Physiopathologie de l'hypertension artérielle pulmonaire : aspects cellulaires et moléculaires
,
2006
.
[6]
S. Glennan.
The modeler in the crib
,
2005
.
[7]
Martin Arlitt,et al.
A workload characterization study of the 1998 World Cup Web site
,
2000,
IEEE Netw..