Managing Web server performance with AutoTune agents

Abstract

Managing the performance of e-commerce sites is challenging. Site content changes frequently, as do customer interests and business plans, contributing to dynamically varying workloads. To maintain good performance, system administrators must tune their information technology environment on an ongoing basis. Unfortunately, doing so requires considerable expertise and increases the total cost of system ownership. In this paper, we propose an agent-based solution that not only automates the ongoing system tuning but also automatically designs an appropriate tuning mechanism for the target system. We illustrate this in the context of managing a Web server. There we study the problem of controlling CPU and memory utilization of an Apache ® Web server using the application-level tuning parameters MaxClients and KeepAlive, which are exposed by the server. Using the AutoTune agent framework under the Agent Building and Learning Environment (ABLE), we construct agents to fully automate a control-theoretic methodology that involves model building, controller design, and run-time feedback control. Specifically, we design (1) a modeling agent that builds a dynamic system model from the controlled server run data, (2) a controller design agent that uses optimal control theory to derive a feedback control algorithm customized to that server, and (3) a run-time control agent that deploys the feedback control algorithm in an on-line realtime environment to automatically manage the Web server. The designed autonomic feedback control system is able to handle the dynamic and interrelated dependencies between the tuning parameters and the performance metrics with guaranteed stability from control theory. The effectiveness of the AutoTune agents is demonstrated through experiments involving variations in workload, server capacity, and business objectives. The results also serve as a validation of the ABLE toolkit and the AutoTune agent framework.The increasing complexity of computing systems and applications demands a correspondingly larger human effort for system configuration and performance management. This manual effort can be timeconsuming and error-prone, and requires highly skilled personnel, making it costly. Autonomic computing 1 uses the analogy of the human autonomic nervous system to suggest the use of a higher level of automation and self-management capability in computing systems.The complexity and importance of developing autonomic computing systems has attracted research

Predictive Autonomicity of Web Services in the MAWeS Framework

“…Then the services of the framework automatically find out the set of simulations needed, perform them (2,3,4), and return the set of optimal parameters for the target application (5). Finally, MHAWeSclient starts the application code, passing to it the set of optimized parameters (6 …”

“…Most of the times, autonomic computing models rely on some sort of feedback control mechanism to provide self-configuration and/or self-optimization capabilities [5] . Following up the seminal study that firstly introduced the use of predictive autonomicity [12] , further work based on feedforward control (or on the joint use of feedforward and feedback) is currently being developed [9] .…”

“…The use of Web Services architectures is becoming a customary approach for the development of open, large-scale interoperable systems [1][2][3][4][5] , and there are many examples of "working" solutions. But reliability, availability, performance and security of these architectures are completely open issues.…”

Self-optimization of secure web services

“…Obtaining optimality for all of the involved parameters for composed services is a challenging and open issue. Optimal performance is related not only to the specific services, but also to the systems that host them and to the way the mapping between service and systems is performed [3][4][5][6][7]. Performance-oriented management of services requires the ability to dynamically change service configuration and in a way transparent to the service offered to the end-user.…”

Toward Broad-Spectrum Autonomic Management

“…Much current research focuses on increasing the performance of a fabric [5], and on improving the security measures deployed upon it. But down time due to configuration errors has a significant effect upon costs [6], [1], while the benefits of modest performance improvements may be harder to quantify.…”