Master Thesis MSTR-0021

BibliographyWaldvogel, Claudio: Specification and Runtime Extraction of Enterprise Application Architectures for Expert-Guided Performance Problem Diagnosis.
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Master Thesis No. 21 (2015).
83 pages, english.
CR-SchemaH.3.4 (Information Storage and Retrieval Systems and Software)
Abstract

The non-functional requirements (NFRs) of enterprise application systems (EASs) have a significant impact on the Key Performance Indicators (KPIs) of companies. Among NFRs like accessibility, security, and reusability is performance considered as one of the top most important. Performance quantifies the degree to which an application meets the requirements, with respect to response times and resource utilization. To enable early performance problem detection, so-called Application Performance Management (APM) tools are integrated in an EAS life cycle.

Due to the high initial and ongoing configuration effort of APM tools, they have hardly been accepted in the industry. This results in time-consuming and error-prone manual performance problem diagnosis. These vulnerabilities of APM tools are addressed by the diagnoseIT research project. The main objective of the project is to enrich existing APM processes with automated configuration of instrumentations as well as automated performance problem detection and diagnosis. Since there is already a wide variety of APM tools, diagnoseIT does not implement a new tool to measure performance metrics. Instead, already existing APM tools provide their monitoring data to diagnoseIT. As part of this research project arose this work and contributed three components to the diagnoseIT framework.

As a basis for performance problem diagnosis, diagnoseIT needs to know a variety of information (e.g., system architecture, execution environment, and dynamic runtime data) about the monitored EAS. Therefore, an Enterprise Performance Model (EPM) was designed and implemented in the first part of this thesis. The second part of the work was to provide a maintenance service for the EPM and an associated integration interface for third-party APM tools. The implemented components were assembled to a prototypical implementation of the diagnoseIT framework. The final evaluation of the implemented solution has shown that we are able to maintain the EPM by connecting the Kieker application performance monitoring tool to diagnoseIT. The evaluation results of extensive load tests showed, however, that the processable amount of data is limited by the current implementation of the persistence unit.

Full text and
other links
PDF (1273104 Bytes)
Department(s)University of Stuttgart, Institute of Software Technology, Software Reliability and Security
Superviser(s)André van Hoorn
Entry dateSeptember 21, 2015
   Publ. Computer Science