Visible to the public Automated and Agile Server ParameterTuning by Coordinated Learning and Control

TitleAutomated and Agile Server ParameterTuning by Coordinated Learning and Control
Publication TypeJournal Article
Year of Publication2014
AuthorsYanfei Guo, Lama, P., Changjun Jiang, Xiaobo Zhou
JournalParallel and Distributed Systems, IEEE Transactions on
Date PublishedApril
Keywordsagile parameter tuning, agile server parameter tuning, automated server parameter tuning, average response time, bursty workloads, cloud environments, control engineering computing, coordinated learning and control, decision making process, effective throughput, fault tolerant computing, fuzzy control, Internet, learning (artificial intelligence), model independence, multitier applications, multitier Internet applications, multitier service architecture, neural fuzzy control, Neural networks, neurocontrollers, Neurons, online learning, parameter tuning direction, predefined adjustment value, quantitative control output, reinforcement learning based server parameter tuning approach, self-adaptiveness, self-adjusting systems, Servers, system throughput, telecommunication computing, Throughput, Time factors, trial-and-error, virtualisation, virtualized data center hosting RUBiS, virtualized server infrastructure, WikiBench benchmark application

Automated server parameter tuning is crucial to performance and availability of Internet applications hosted in cloud environments. It is challenging due to high dynamics and burstiness of workloads, multi-tier service architecture, and virtualized server infrastructure. In this paper, we investigate automated and agile server parameter tuning for maximizing effective throughput of multi-tier Internet applications. A recent study proposed a reinforcement learning based server parameter tuning approach for minimizing average response time of multi-tier applications. Reinforcement learning is a decision making process determining the parameter tuning direction based on trial-and-error, instead of quantitative values for agile parameter tuning. It relies on a predefined adjustment value for each tuning action. However it is nontrivial or even infeasible to find an optimal value under highly dynamic and bursty workloads. We design a neural fuzzy control based approach that combines the strengths of fast online learning and self-adaptiveness of neural networks and fuzzy control. Due to the model independence, it is robust to highly dynamic and bursty workloads. It is agile in server parameter tuning due to its quantitative control outputs. We implemented the new approach on a testbed of virtualized data center hosting RUBiS and WikiBench benchmark applications. Experimental results demonstrate that the new approach significantly outperforms the reinforcement learning based approach for both improving effective system throughput and minimizing average response time.

Citation Key6497051