Download presentation
Presentation is loading. Please wait.
Published byCody Martin Modified over 9 years ago
1
1 Economical and Robust Provisioning of N-Tier Cloud Workloads: A Multi-level Control Approach Pengcheng Xiong 1, Zhikui Wang 2, Simon Malkowski 1, Qingyang Wang 1, Deepal Jayasinghe 1, Calton Pu 1 1 Georgia Institute of Technology 2 HP Labs Email: xiong@gatech.edu
2
Overview Motivation Background Resource partition controller Application controller Conclusions 2
3
Overview Motivation Background Resource partition controller Application controller Conclusions 3
4
Applications in a typical Cloud environment
5
Different feedback controller design for a single/multi-tiered application (1) 5 Zhu et al, ACC 2006
6
Different feedback controller design for a single/multi-tiered application (2) 6 Wang et al, FeBID 2007 T UC T FB T FF
7
Different controllability under different workload generator (1) 7 Schroeder et al, NSDI 2006
8
Different controllability under different workload generator (2) 8 Xiong et al, NOMS 2010
9
Goals Economical –We want to meet the performance requirement for the N-tier web application with the minimum total resources. Robust –We want to be robust to different time-varying workload types, e.g., open, closed, semi-open. 9
10
Overview Motivation Background Resource partition controller Application controller Conclusions 10
11
Control Architecture 11
12
Test bed Experiment Environment –Apache, Tomcat, Mysql –Xen hypervisor Workload Generator –RUBiS “Browsing mix” workload that has 10 transaction types, e.g., Home, Browse, ViewItem. (just like eBay) –Workload types (open, closed, semi-open) –Workload intensity 12
13
Overview Motivation Background Resource partition controller Application controller Conclusions 13
14
System modeling 14
15
Optimal resource partition Solution 1(Shares) Solution 2(Util.) Our solution(Opt.)
16
Evaluation of resource partition controller 16
17
Overview Motivation Background Resource partition controller Application controller Conclusions 17
18
Application controller design 18 System model between the RTT and S –System identification method based on ARMA model Controller design –Root-locus method based on control theory
19
System identification 19
20
Controller design 20 ARX01 model Proportional-integral (PI) controller The closed model transfer function
21
Performance controller(setting=35ms) 21 Util has MORE fluctuation than Opt.
22
Performance controller(setting=200ms) 22
23
Conclusions We propose economical and robust provisioning for Cloud resources for N-tier web applications through a multi-level control approach. Experimental results show that our solution outperforms other existing approaches –Almost the same performance but save up to 20% CPU resources. –Robust to deal with different workload styles. 23
24
24 Thanks
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.