Autor*innen mit den meisten Dokumenten
Auflistung nach:
Neueste Veröffentlichungen
- KonferenzbeitragTowards Bringing Vitruvius into the Cloud(Softwaretechnik-Trends Band 44, Heft 4, 2024) Armbruster, Martin; Weber, Thomas; König, LarsModel consistency, e.g., to keep an architectural model and a source code model of a system consistent, is a topic of active research, driven by a need from industry. Academic approaches, e.g., Vitruvius, provide solutions to such pressing problems, but often lack usability, e.g., for collaborative modeling, which is a use case we want to provide in the future. To improve the usability of Vitruvius, we extended it with server client functionality. This improves the usability, but may introduce performance disadvantages. Thus, we investigated and compared the performance of our new server-client implementation to local instances, which are monolithic, to estimate the additional overhead. The results indicate that the server-client implementation can add minimal overhead.
- KonferenzbeitragAutomatic Performance Modeling of Configurable Scientific Software(Softwaretechnik-Trends Band 44, Heft 4, 2024) Schmid, LarissaModern software is configurable and allows users to set many parameters according to their needs. Due to many non-functional parameters, usually, the same functionality can be achieved with varying performance. Performance models express application performance as functions of input parameters, helping users and developers understand application behavior. Automatic performance modeling approaches can generate performance models automatically from empirical measurements of the software. Current modeling approaches employ heuristics for deciding which configurations to measure, resulting in a trade-off between the cost of measurements and accuracy of the model. To overcome this trade-off, we propose approaches to derive the smallest necessary measurement setup based on results of a system analysis, and to automatically identify performance-irrelevant options. Our evaluation with real-world applications show that we can significantly decrease cost of performance modeling while maintaining accuracy of the resulting models.
- KonferenzbeitragOverhead Measurement Noise in Different Runtime Environments(Softwaretechnik-Trends Band 44, Heft 4, 2024) Reichelt, David Georg; Jung, Reiner; van Hoorn, AndréIn order to detect performance changes, measurements are performed with the same execution environment. In cloud environments, the noise from different processes running on the same cluster nodes might change measurement results and thereby make performance changes hard to measure. The benchmark MooBench determines the overhead of different observability tools and is executed continuously. In this study, we compare the suitability of different execution environments to benchmark the observability overhead using MooBench. To do so, we compare the execution times and standard deviation of MooBench in a cloud execution environment to three bare-metal execution environments. We find that bare metal servers have lower runtime and standard deviation for multi-threaded MooBench execution. Nevertheless, we see that performance changes up to 4.41 % are detectable by GitHub actions, as long as only sequential workloads are examined.
- KonferenzbeitragEvaluating the Overhead of the Performance Profiler Cloudprofiler With MooBench(Softwaretechnik-Trends Band 44, Heft 4, 2024) Yang, Shinhyung; Reichelt, David Georg; Hasselbring, WilhelmPerformance engineering has become crucial for the cloud-native architecture. This architecture deploys multiple services, with each service representing an orchestration of containerized processes. OpenTelemetry is growing popular in the cloud-native industry for observing the software’s behaviour, and Kieker provides the necessary tools to monitor and analyze the performance of target architectures. Observability overhead is an important aspect of performance engineering and MooBench is designed to compare different observability frameworks, including OpenTelemetry and Kieker. In this work, we measure the overhead of Cloudprofiler, a performance profiler implemented in C++ to measure native and JVM processes. It minimizes the profiling overhead by locating the profiler process outside the target process and moving the disk writing overhead off the critical path with buffer blocks and compression threads. Using MooBench, Cloudprofiler’s buffered ID handler with the Zstandard lossless data compression ZSTD showed an average execution time of 2.28 microseconds. It is 6.15 times faster than the non-buffered and non-compression handler.
- KonferenzbeitragThe Hidden Costs of Shared CPU Resources: A Closer Look at Cgroups and QoS(Softwaretechnik-Trends Band 44, Heft 4, 2024) Volpert, Simon; Winkelhofer, Sascha; Seybold, Daniel; Domaschka, Jörg; Wesner, StefanPrevious research has demonstrated that Cgroups CPU isolation, while effective in guaranteeing resource limits, may not fully prevent performance degradation in co-located applications. This paper investigates the underlying causes of this performance drop by examining the impact of Hyper-Threading (HT) and Dynamic Frequency Scaling (DFS). Through controlled experiments using a database workload and a CPU intensive stressor, we analyze the changes in Quality of Service (QoS) isolation when these features are disabled. Our findings reveal the degree to which these mechanisms contribute to the observed performance interference and provide insights into the limitations of relying solely on Cgroups for robust CPU QoS isolation.
- KonferenzbeitragAn Empirical Study on the Impact of Selected Host Configuration Parameters on Container Start Times(Softwaretechnik-Trends Band 44, Heft 4, 2024) Straesser, Martin; Erhard, Nicholas; Kounev, SamuelModern cloud technologies rely on container virtualization as the basis for application deployment as they show faster start times compared to virtual machines. Understanding the impacting factors for container start times is crucial for application developers and platform engineers to identify optimization potential. In this work, we extend our previous empirical study on container start times by analyzing the impact of selected host configuration parameters (e.g., disk type, RAM size, CPU vendor). Our insights include that the disk type is the most important platform parameter analyzed. However, attributes of the container image and the host machine must always be considered together to assess container start times accurately.
- Konferenz-AbstractSoftware Engineering 2025: Aufruf zur Einreichung von Beiträgen(Softwaretechnik-Trends Band 44, Heft 4, 2024) Koziolek, Anne; Alpers, Sascha; Lamprecht, Anna-Lena; Thüm, Thomas; Hey, Tobias; Burger, ErikSoftware Engineering 2025: Aufruf zur Einreichung von Beiträgen
- Konferenz-AbstractCall for Papers: WSRE 2025 - 27. Workshop Software-Reengineering & -Evolution(Softwaretechnik-Trends Band 44, Heft 4, 2024) Quante, Jochen; Konersmann, Marco; Sauer, Stefan; Schilling, Daniela; Schulze, SandroCall for Papers: WSRE 2025 - 27. Workshop Software-Reengineering & -Evolution
- Konferenz-AbstractInformatik Festival 2025: Jetzt Workshops einreichen(Softwaretechnik-Trends Band 44, Heft 4, 2024) Scheibe, Alexander; Durst, MarkusInformatik Festival 2025: Call for Workshops
- Konferenz-Abstract15th Symposium on Software Performance 2024. Linz, Austria, November 6–7, 2024(Softwaretechnik-Trends Band 44, Heft 4, 2024) Henning, Sören; Kahlhofer, Mario; Vogel, AdrianoThe 15th edition of the Symposium on Software Performance, held November 6–7, 2024, in Linz, Austria, brought together researchers and practitioners with a shared interest in all aspects of software performance. This edition attracted over 50 participants from 16 affiliations, including well-known educational and research institutions from Austria, Germany, and Denmark, as well as prominent industry companies.