Performance Problem Diagnostics By Systematic Experimentation

In this book, we introduce an automatic, experiment-based approach for performance problem diagnostics in enterprise software systems. The proposed approach systematically searches for root causes of detected performance problems by executing series of systematic performance tests. The presented approach is evaluated by various case studies showing that the presented approach is applicable to a wide range of contexts. This work was published by Saint Philip Street Press pursuant to a Creative Commons license permitting commercial use. All rights not granted by the work's license are retained by the author or authors.

In this book, we introduce an automatic, experiment-based approach for performance problem diagnostics in enterprise software systems. The proposed approach systematically searches for root causes of detected performance problems by executing series of systematic performance tests. The presented approach is evaluated by various case studies showing that the presented approach is applicable to a wide range of contexts.

Performance modelling can require substantial effort when creating and maintaining performance models for software systems that are based on existing software. Therefore, this thesis addresses the challenge of performance prediction in such scenarios. It proposes a novel goal-oriented method for experimental, measurement-based performance modelling. We validated the approach in a number of case studies including standard industry benchmarks as well as a real development scenario at SAP.

The quality of enterprise software applications plays a crucial role for the satisfaction of the users and the economic success of the enterprises. Software applications with unsatisfying performance and scalability are perceived by its users as low in quality, as less interesting and less attractive, and cause frustration when preventing the users from attaining their goals. This book proposes an approach for a recommendation system that enables developers who are novices in software perform.

This work developed an automatic approach for the assessment of software reliability which is both theoretical sound and practical. The developed approach extends and combines theoretical sound approaches in a novel manner to systematically reduce the overhead of reliability assessment.

This book provides an overview of the problems involved in engineering scalable, elastic, and cost-efficient cloud computing services and describes the CloudScale method — a description of rescuing tools and the required steps to exploit these tools. It allows readers to analyze the scalability problem in detail and identify scalability anti-patterns and bottlenecks within an application. With the CloudScale method, software architects can analyze both existing and planned IT services. The method allows readers to answer questions like: • With an increasing number of users, can my service still deliver acceptable quality of service? • What if each user uses the service more intensively? Can my service still handle it with acceptable quality of service? • What if the number of users suddenly increases? Will my service still be able to handle it? • Will my service be cost-efficient? First the book addresses the importance of scalability, elasticity, and cost-efficiency as vital quality-related attributes of modern cloud computing applications. Following a brief overview of CloudScale, cloud computing applications are then introduced in detail and the aspects that need to be captured in models of such applications are discussed. In CloudScale, these aspects are captured in instances of the ScaleDL modeling language. Subsequently, the book describes the forward engineering part of CloudScale, which is applicable when developing a new service. It also outlines the reverse and reengineering parts of CloudScale, which come into play when an existing (legacy) service is modified. Lastly, the book directly focuses on the needs of both business-oriented and technical managers by providing guidance on all steps of implementing CloudScale as well as making decisions during that implementation. The demonstrators and reference projects described serve as a valuable starting point for learning from experience. This book is meant for all stakeholders interested in delivering scalable, elastic, and cost-efficient cloud computing applications: managers, product owners, software architects and developers alike. With this book, they can both see the overall picture as well as dive into issues of particular interest.

Die modellbasierte Performancevorhersage ist ein bekanntes Konzept zur Gewährleistung der Softwarequalität. Derzeitige Ansätze basieren auf einem Modell mit einer Metrik, was zu ungenauen Vorhersagen für moderne Architekturen führt. In dieser Arbeit wird ein Multi-Strategie-Ansatz zur Erweiterung von Performancevorhersagemodellen zur Unterstützung von Multicore-Architekturen vorgestellt, in Palladio implementiert und dadurch die Genauigkeit der Vorhersage deutlich verbessert. - Model-based performance prediction is a well-known concept to ensure the quality of software. Current approaches are based on a single-metric model, which leads to inaccurate predictions for modern architectures. This thesis presents a multi-strategies approach to extend performance prediction models to support multicore architectures. We implemented the strategies into Palladio and significantly increased the performance prediction power.