Browsing by Author "Angerstein, Tobias"

Now showing 1 - 3 of 3

Open Access
Automated categorization of performance problem diagnosis results
(2016) Angerstein, Tobias
In times of big data and global networking, software performance becomes a major issue in software development. Many enterprise applications were not designed for the current highly frequented usage. Thus, more and more organizations are using Application Performance Monitoring tools, to detect bottlenecks and other performance problems in their product. Meanwhile, there are many different tools on the market which provide a detailed performance-aware analysis of enterprise applications. However the common Application Performance Management tools do not provide any additional automated performance problem diagnosis. The performance expert has to find the cause on its own. diagnoseIT [4], a framework for automatic diagnosis of performance problems in enterprise applications addresses this problem. diagnoseIT extracts performance problem instances from execution traces. The resulting set of problem instances can become huge and the analysis of the problems is very time-consuming. We extend the general concept of diagnoseIT and categorize the resulting set of problem instances into a manageable number of problem categories. Therefore, a concept of categorization is elaborated. We analyze several different categorization approaches and evaluate the performance and the quality of the result. We perform a sensitivity analysis, which analyzes the influence of each attribute of a problem instance on a clustering result. The results of the sensitivity analysis indicates, that there is a potential for optimization. Thus, we introduce a concept of optimization, which optimizes the process of categorization by weighting the attributes and we compare different manual and automatic optimization approaches with regard to the improvement compared to default weights. In the evaluation, we examine the accuracy and the performance of the approaches. The evaluation shows that k-means clustering provides the most promising and best results. Additionally, the evaluation indicates a high potential for optimization. However, the results of the evaluation show that it is difficult to optimize the weights without any knowledge about the analyzed system.
Open Access
ContinuITy - automatisiertes Performance-Testen in der kontinuierlichen Softwareentwicklung : Abschlussbericht
(2020) Angerstein, Tobias; Heger, Christoph; Hoorn, André van; Okanović, Dušan; Schulz, Henning; Siegl, Stefan; Wert, Alexander
Ziel des Forschungsprojekts ContinuITy war die Entwicklung eines Ansatzes und entsprechender Werkzeugunterstützung zum automatisierten Performance-Testen - eingebettet in Prozesse und Infrastruktur der kontinuierlichen Softwareentwicklung. Ziel ist durch Ausnutzung von kontinuierlich aufgezeichneten Messdaten aus dem Produktivbetrieb, automatisiert effizientes und nachhaltiges Lasttesten zu gewährleisten und in die kontinuierliche Softwareentwicklung zu integrieren. Lasttests werden automatisiert aus Messdaten extrahiert und evolviert. Lasttests werden durch eine modulare Beschreibungssprache definiert und können durch zusätzliche Semantik - z. B. über Testart und -ziele - angereichert werden. Im Rahmen der Automatisierung des Softwareerstellungsprozesses (Continuous Delivery) erfolgt eine Auswahl relevanter Lasttests, die Erkennung von Regressionen und deren Diagnose. Basierend auf der Beschreibung der Aufgabenstellung und der Voraussetzungen fasst dieser Bericht die durchgeführten Aktivitäten und wesentlichen Ergebnisse zusammen.
Open Access
Modularization of representative load tests for microservice applications
(2018) Angerstein, Tobias
Nowadays, user satisfaction and business value of software applications are mainly influenced by the performance and scalability of software applications. Thus, it is a necessity to test the performance of the application during the application development cycle. In order to have meaningful results, the whole application has to be tested with a representative load. Using representative load tests, the application’s performance is tested by a simulated request load, which represents a realistic group of users. In times of DevOps and microservice architectures, each microservice is developed, tested and deployed by a different development team. However, existing approaches, which provide representative load testing, only generate workload for the whole system. Thus, all services have to be deployed, have to be configured by hand, and the execution takes very long compared to typical delivery pipeline executions. In addition, this contradicts the idea of microservice architecture development, where each team develops, tests and deploys its service independently. Addressing the downsides of system-wide load testing regarding microservices, we propose a concept of a representative load test workload modularization. Instead of targeting the whole system, only dedicated services are targeted by modularizing the workload into service-specific parts. Providing modularized load tests for certain services saves many resources and we expect, that it eases the test integration into a continuous delivery pipeline. This thesis aims to achieve the following goals: • Automated representative per-service and integration performance testing • Simplify the use of load testing for microservice development teams Since representative load testing requires a pipeline starting from extracting monitoring data, processing the data, generating a workload model and transforming the workload model into an executable load test, there are many different stages, were the modularization of a load test can take place. In this thesis, we elaborate on possible modularization approaches and evaluate them regarding the resource consumption, duration, representativeness, and the practicability. We implement two of the elaborated approaches, in order to evaluate and compare the results of the modularization approaches. A selection of microservices, which are part of a sample application is load tested with the implemented approaches and compared with the non-modularized load test executions. The results of the experiment show, that the trace modularization approach, which modularizes the request traces, provides the most promising results. However, the results are not significant enough in order to draw a valid conclusion.