ABSTRACT
Elastic n-tier applications have non-stationary workloads that require adaptive control of resources allocated to them. This presents not only an opportunity in pay-as-you-use clouds, but also a challenge to dynamically allocate virtual machines appropriately. Previous approaches based on control theory, queuing networks, and machine learning work well for some situations, but each model has its own limitations due to inaccuracies in performance prediction. In this paper we propose a multi-model controller, which integrates adaptation decisions from several models, choosing the best. The focus of our work is an empirical model, based on detailed measurement data from previous application runs. The main advantage of the empirical model is that it returns high quality performance predictions based on measured data. For new application scenarios, we use other models or heuristics as a starting point, and all performance data are continuously incorporated into the empirical model's knowledge base. Using a prototype implementation of the multi-model controller, a cloud testbed, and an n-tier benchmark (RUBBoS), we evaluated and validated the advantages of the empirical model. For example, measured data show that it is more effective to add two nodes as a group, one for each tier, when two tiers approach saturation simultaneously.
- Animoto's Facebook Scale-up. http://blog.rightscale.com/2008/04/23/animoto-facebook-scale-up/, 2008.Google Scholar
- M. Arlitt and T. Jin: A workload characterization study of the 1998 world cup web site. Network '00.Google ScholarDigital Library
- M. Armbrust, A. Fox, D. A. Patterson, N. Lanham, B. Trushkowsky, et al.: SCADS: Scale-independent storage for social computing applications. CIDR '09.Google Scholar
- I. Cohen, M. Goldszmidt, et al.: Correlating instrumentation data to system states: a building block for automated diagnosis and control. OSDI '04. Google ScholarDigital Library
- I. Cohen, S. Zhang, et al.: Capturing, indexing, clustering, and retrieving system history. SOSP '05. Google ScholarDigital Library
- D. Feitelson: Workload modeling for computer systems performance evaluation. http://www.cs.huji.ac.il/~feit/wlmod/, 2011.Google Scholar
- M. Hedwig, S. Malkowski, and D. Neumann: Taming energy costs of large enterprise systems through adaptive provisioning. ICIS '09.Google Scholar
- M. Hedwig, S. Malkowski, et al.: Towards autonomic cost-aware allocation of cloud resources. ICIS '10.Google Scholar
- R. Jain: The art of computer systems performance analysis. John Wiley & Sons, Inc., 1991.Google Scholar
- G. Jung, K. Joshi, M. Hiltunen, et al.: Generating adaptation policies for multi-tier applications in consolidated server environments. ICAC '08. Google ScholarDigital Library
- G. Jung, K. R. Joshi, M. A. Hiltunen, et al.: A cost-sensitive adaptation engine for server consolidation of multitier applications. Middleware '09. Google ScholarDigital Library
- H. C. Lim, S. Babu, and J. S. Chase: Automated control for elastic storage. ICAC '10. Google ScholarDigital Library
- S. Malkowski, M. Hedwig, D. Jayasinghe, C. Pu, and D. Neumann: CloudXplor: A tool for configuration planning in clouds based on empirical data. SAC '10. Google ScholarDigital Library
- S. Malkowski, M. Hedwig, and C. Pu: Experimental evaluation of N-tier systems: Observation and analysis of multi-bottlenecks. IISWC '09. Google ScholarDigital Library
- S. Malkowski, D. Jayasinghe, et al.: %M. Hedwig, J. Park, Y. Kanemasa, and C. Pu. Empirical analysis of database server scalability using an n-tier benchmark with read-intensive workload. SAC '10. Google ScholarDigital Library
- N. Mi, G. Casale, L. Cherkasova, and E. Smirni: Burstiness in multi-tier applications: symptoms, causes, and new models. Middleware '08. Google ScholarDigital Library
- N. Mi, G. Casale, L. Cherkasova, and E. Smirni: Injecting realistic burstiness to a traditional client-server benchmark. ICAC '09. Google ScholarDigital Library
- P. Padala, K.-Y. Hou, et al.: Automated control of multiple virtualized resources. EuroSys '09. Google ScholarDigital Library
- P. Padala, K. G. Shin, X. Zhu, M. Uysal, Z. Wang, et al.: Adaptive control of virtualized resources in utility computing environments. EuroSys '07. Google ScholarDigital Library
- A. Sahai, S. Singhal, and Y. B. Udupi: A classification-based approach to policy refinement. IM '07.Google Scholar
- C. Stewart, T. Kelly, et al.: %and A. Zhang: Exploiting nonstationarity for performance prediction. EuroSys '07. Google ScholarDigital Library
- RUBBoS: Bulletin board benchmark. http://jmob.objectweb.org/rubbos.html, 2008.Google Scholar
- E. Thereska and G. R. Ganger: IRONModel: robust performance models in the wild. SIGMETRICS '08. Google ScholarDigital Library
- B. Urgaonkar, G. Pacifici, P. Shenoy, M. Spreitzer, et al.: An analytical model for multi-tier internet services and its applications. SIGMETRICS '05. Google ScholarDigital Library
- B. Urgaonkar, P. Shenoy, et al.: Dynamic provisioning of multi-tier internet applications. ICAC '05. Google ScholarDigital Library
- P. Xiong, Y. Chi, S. Zhu, H. J. Moon, C. Pu, et al.: Intelligent management of virtualized resources for database systems in cloud environment. ICDE '11. Google ScholarDigital Library
- P. Xiong, Z. Wang, G. Jung, and C. Pu: Study on performance management and application behavior in virtualized environment. NOMS '10.Google Scholar
Index Terms
- Automated control for elastic n-tier workloads based on empirical modeling
Recommendations
Automated control in cloud computing: challenges and opportunities
ACDC '09: Proceedings of the 1st workshop on Automated control for datacenters and cloudsWith advances in virtualization technology, virtual machine services offered by cloud utility providers are becoming increasingly powerful, anchoring the ecosystem of cloud services. Virtual computing services are attractive in part because they enable ...
Automated control of multiple virtualized resources
EuroSys '09: Proceedings of the 4th ACM European conference on Computer systemsVirtualized data centers enable sharing of resources among hosted applications. However, it is difficult to satisfy service-level objectives(SLOs) of applications on shared infrastructure, as application workloads and resource consumption patterns ...
Automated control for elastic storage
ICAC '10: Proceedings of the 7th international conference on Autonomic computingElasticity - where systems acquire and release resources in response to dynamic workloads, while paying only for what they need - is a driving property of cloud computing. At the core of any elastic system is an automated controller. This paper ...
Comments