Informationen zu Vorträgen und Referenten des KoSSE-Symposium Application Performance Management (Kieker Days 2012).
Donnerstag, 29. Nov 2012
Johannes Weigend (QAware, München): Dynamische Analyse mit dem Software-EKG, 12:15–12:55 Uhr
Der Vortrag zeigt, wie man komplexe, heterogene Systeme analysiert, wenn die einfachen Methoden (Logfiles, Debugger, Profiler) nicht ausreichen. Ziel der Analysen ist die Ursachenermittlung von Stabilitäts- und Perfomanzproblemen.
Der Vortrag erläutert die Grundlagen, beschreibt ein Vorgehen mit den nötigen Tools und bringt einige Beispiele aus der Praxis.
Biographie
Johannes Weigend ist Chefarchitekt, Geschäftsführer und Mitgründer der QAware. Er studierte Informatik mit Schwerpunkt “Verteilte Systeme” an der Hochschule Rosenheim und hält dort Vorlesungen seit 2001. Er war Leiter der Schulungseinheit Software Engineering bei IXOS, München. Bei der QAware verantwortet Johannes Weigend Forschung & Entwicklung und die technische Infrastruktur. Er ist zuständig für den Kunden BMW.
André van Hoorn (CAU Kiel): Kieker: Overview, Review, and Outlook, 12:55–13:20 Uhr
Application performance monitoring and dynamic analysis of software systems are the basis for various quality-of-service management and reverse-engineering activities and approaches. In this area, Kieker has evolved from a basic AOP-based Java method response time logger to a mature and extensible framework providing monitoring, analysis, and visualization support, employed for many different uses in research, teaching, and practice. Starting with an overview about the framework, this talk will review the past years of Kieker development, before giving an outlook over future directions.
Biographie
André van Hoorn is researcher and PhD student with the Software Engineering Group at Kiel University. In 2007, he received his Master’s (Diploma) degreee in Computer Science from the University of Oldenburg. His research interests focus on architecture-based and model-driven approaches for software performance engineering, self-adaptation, and re(verse)-engineering based on dynamic software analysis. Together with a couple of colleagues, André is developing the Kieker framework (kieker-monitoring.net).
Christoph Heger und Alexander Wert (KIT & SAP Research, Karlsruhe): Systematic Experimentation with Kieker, 13:20–14:00 Uhr
Performance problems pose a significant risk to software vendors. If left undetected, they can lead to lost customers, increased operational costs, and damaged reputation. Despite all efforts, software engineers cannot fully prevent performance problems being introduced into an application. Detecting and resolving such problems as early as possible with minimal effort is still an open challenge in software performance engineering. To address this challenge, we present two approaches for performance problem identification which are based on systematic experimentation. The Kieker-Framework is primarily designed for run-time monitoring rather than for systematic experimentation. Still, we utilize Kieker for systematic experimentation as it provides a sophisticated infrastructure for gathering measurement data of different kind across the whole application under test. With our two approaches, we introduce a new application scenario for Kieker and present our experience using Kieker for systematic experimentation.
Biographie C. Heger
Christoph Heger is a Ph.D. student with the Software Design and Quality Group at the Karlsruhe Institute of Technology (KIT) delegated to the Lean Performance Engineering Group at SAP Research in Karlsruhe. From 2006 to 2012, he studied Computer Science at the KIT and received the M.Sc. (Diplom) in February 2012. His research interests include software performance engineering and software design. In particular, Christoph is interested in providing feedback on performance and scalability for data-centric applications.
Biographie A. Wert
Alexander Wert is a Ph.D. student with the Software Design and Quality Group at the Karlsruhe Institute of Technology (KIT) delegated to the Lean Performance Engineering Group at SAP Research in Karlsruhe. From 2006 to 2012, he studied Computer Science at the KIT and received the M.Sc. (Diplom) in April 2012. His research interests include software performance engineering and software design. In particular, Alexander is interested in measurement-based approaches for performance problem detection and solution.
Holger Knoche (b+m Informatik AG, Melsdorf): Use-Case-basierte Abhängigkeitsanalyse von Legacy-Systemen mit Kieker, 14:20–15:00 Uhr
In vielen Softwaresystemen ist nicht mehr bekannt, welche Codeteile an der Abarbeitung welcher Use-Cases beteiligt sind und wie diese interagieren. Insbesondere im Kontext der Modernisierung von Legacy-Systemen ist dieses Wissen aber von großer Bedeutung. Dieser Vortrag zeigt, wie diese Informationen durch eine Kombination von statischen und dynamischen Analysetechniken wiedergewonnen werden können und stellt die zu diesem Zweck entwickelten neuen Features von Kieker vor. Als Fallstudien dienen zwei industrielle Softwaresysteme, die in COBOL bzw. Visual Basic 6 implementiert sind.
Biographie
Holger Knoche ist seit 2005 als Softwareentwickler und -architekt bei der b+m Informatik AG tätig und verfügt über langjährige Erfahrung aus Softwareprojekten, insbesondere im Finanzdienstleistungsbereich. Derzeit beschäftigt er sich als Senior-Softwarearchitekt bei b+m u.a. mit modellgetriebener Modernisierung.
Ana Dragomir (RWTH Aachen): Model-based Architecture Evolution and Evaluation, 15:00–15:40 Uhr
Software systems tend to evolve independently from their architecture description. Within the ARAMIS project, we are developing an approach to continuously monitor and evaluate the state of the architecture and to support the goal-based evolution of the software system. The monitoring will be based on the Kieker framework and will produce up-to-date architecture descriptions at various abstraction levels and from various viewpoints. The creation of rules that should be imposed on the software architecture is also part of our research. These rules will be used to periodically check if the architecture is still conformant with them. Existing or newly developed metrics will further be used to regularly evaluate the state of the architecture, thus offering a continuous overview of the quality evolution of the architecture. Finally, we intend to develop a method to define architecture variants and to analyze which of them
better sustains a reasonable software evolution.
Biographie
Ana Dragomir is a Ph.D. student and research assistant with the Software
Construction Group at the RWTH Aachen University. She has finished a Masters in Software Systems Engineering at the RWTH Aachen University after having attended a Bachelor in Computer Science at the Politehnica University Bucharest. Her research interests include software architecture design and analysis, architecture documentation, architecture enforcement as well as architecture knowledge-management.
Peer Brauer (CAU Kiel): Kieker.Workflow: Workflow-Monitoring with Kieker, 15:40–16:10 Uhr
Kieker.Workflow is a workflow monitoring extension to the Kieker framework. It allows Kieker to monitor BPEL workflows, which are executed within the Apache ODE workflow engine. Kieker.Workflow adds several new parts to the Kieker framework, like a new buffered probe, a monitoring information store based on a graph database and a graphical viewer for workflow-related monitoring information. The extension was developed in the context of the PubFlow project, which is about using workflows to enhance the process of data publication.
Biographie
Peer Brauer is a Ph.D. student and researcher with the Software Engineering Group at Kiel University since 2011. He received his diploma in computer science from Kiel University in 2009. From 2009 to 2011 he worked as an IT Consultant in different IT projects. Peer’s research interests include workflow monitoring, data management, and data provenance.
Tillmann Bielefeld (empuxa GmbH, Kiel): OPAD: Online Performance Anomaly Detection with Kieker, 16:30–17:00 Uhr
Provisioning satisfying quality of service (QoS) is a challenge when operating large-scale software systems. Solutions for monitoring QoS metrics like performance and availability do exist and manual problem detection and analysis based on the collected data is common practice. In interactive scenarios, however, users notice anomalies immediately and potential problems should be detected automatically. For large-scale systems, this is hard to achieve, since they typically operate in heterogeneous and evolving environments with system-specific measures and characteristics.
Approaching these challenges, we developed the Kieker-based OPAD tool for online performance anomaly detection. Different time series algorithms can be configured and evaluated in order to address system-specific characteristics. With the use of selected algorithms, it can detect and signal anomalies online and additionally store them for post-mortem analyses. The social network system XING served as a case study and the evaluation of OPAD in this production environment shows promising results in terms of robustness and accuracy.
Biographie
Tillmann Bielefeld hat sein Informatik-Studium an der Universität Kiel im Frühjahr 2012 erfolgreich abgeschlossen. Seine Diplomarbeit über Performance-Anomalieerkennung schrieb er in Kooperation mit der XING AG in Hamburg. Inzwischen widmet sich Till seinem Joint-Venture-Unternehmen (Berlin) für erneuerbare Energien mit dem Schwerpunkt Softwareentwicklung für Energieeffizenz, Visualisierung und Vermarktung von Alternativ-Strom. Bereits während seines Studiums gründete er in Kiel die Firma empuxa GmbH für die Entwicklung von Web und Mobile-Applications. Sein Faible für Softwareprojekte, Sprachen und Kitesurfen brachte ihn zu verschiedenen Firmen, Länder und Kulturen.
Nikolas Herbst (KIT Karlsruhe): Workload Classification and Forecasting, 17:00–17:40 Uhr
As modern enterprise software systems become increasingly dynamic, workload forecasting techniques are gaining in importance as a foundation for onlinecapacity planning and resource management. Time series analysis offers a broad spectrum of methods to calculate workload forecasts based on history monitoring data.
We propose a novel self-adaptive approach that selects suitable forecasting methods for a given context based on a decision tree and direct feedback cycles together with a corresponding implementation. The user needs to provide only his general forecasting objectives. In several experiments based on real-world workload traces, we show that our implementation of the approach provides continuous and reliable forecast results at run-time with an improved accuracy. In a case study, between 55% and 75% of the violations of a given service level agreement can be prevented by applying proactive resource provisioning based on the forecast results of our implementation.
Biographie
This September, Nikolas Roman Herbst graduated at the Faculty of Informatics of the Karlsruhe Institute of Technology and is starting his Ph.D studies at the Software Quality Departement led by Prof. Dr. Ralf Reussner. His diploma thesis on workload classification and forecasting was conducted at the Descartes Research Group led by Dr. Samuel Kounev in cooperation with the Christian-Albrecht University in Kiel and IBM Research and Development in Boeblingen. In particular, he is interested in techniques for workload characterisation and forecasting as well as in metrics and means to benchmark elasticity of virtualized ressources.
Wilhelm Hasselbring (CAU Kiel): iObserve: Integrated Observation and Modeling Techniques to Support Adaptation and Evolution of Software Systems, 17:40–17:55 Uhr
The new DFG Priority Program (SPP1593) on “Design For Future – Managed Software Evolution” has been established to develop fundamentally new approaches in software engineering with a determined focus on long-living software systems.
The increased adoption of service-oriented technologies and cloud computing creates new challenges for the adaptation and evolution of long-living software systems. Software services and cloud platforms are owned and maintained by independent parties. Software engineers and system operators of long-living software systems only have limited visibility and control over those third-party elements. Traditional monitoring provides software engineers and system operators with execution observation data which are used as basis to detect anomalies. If the services and the cloud platform are not owned and controlled by the engineers of the software systems, monitoring the execution of the software system is not straightforward.
As part of the DFG Priority Program, the project iObserve will develop and validate new advanced techniques which empower the system engineers to observe and detect anomalies of the execution of software systems they do not fully own and control. It will extend and integrate previous work on adaptive monitoring, online testing and benchmarking and will use models@runtime as means to adjust the observation and anomaly detection techniques during system operation.
Biographie
Prof. Dr. Wilhelm (Willi) Hasselbring is head of the Software Engineering Group at Kiel University, Dean of the Faculty of Engineering at University of Kiel, Scientific head of the User Group Softwarearchitektur at Softwareforen Leipzig, Member of the SPEC Research Group Steering Committee, Spokesperson of the Kiel KoSSE projects and Principal Investigator of the excellence cluster “Future Ocean” (Second Phase).
Holger Eichelberger (Uni Hildesheim): SPASS-meter – Measuring Diverse Software Attributes in an Integrated Manner, 17:55–18:10 Uhr
SPASS-meter is a novel framework for monitoring software attributes which indicate the resource consumption of a Java program at runtime.
Basically, SPASS-meter measures CPU-time consumption, response time, memory usage, file and network transfer. Further, SPASS-meter supports the aggregation of these attributes at various levels including program and operating system levels. In particular, SPASS-meter supports the flexible aggregation of the monitored software attributes to user-defined logical program units such as components or services, i.e. SPASS-meter enables the user to focus on the relevant software attributes for specific parts of a program. Currently, SPASS-meter can be applied to traditional Java programs as well as to Android Apps. In this talk, we present the concepts of SPASS-meter as well as initial application results.
Biographie
Holger Eichelberger holds a PhD since 2005 and a Dipl.-Inf. in computer science from the University of Würzburg since 1999. His specific interests are adaptive systems (in combination with product line engineering and model based software engineering) as well as programming languages. Currently, he works in the EU-funded project INDENICA on customizing service based systems and in the national funded project ScaleLog on customizing software systems in the logistics domain.
Freitag, 30. Nov 2012
Lennart Koopmann (XING AG, Hamburg): Log-Management mit Graylog2, 9:00–9:40 Uhr
Das positive zuerst: Viele Unternehmen speichern ihre Logs inzwischen zentral. Leider werden die Möglichkeiten, die in diesen Logs liegen sehr häufig nicht genutzt. In diesem Vortrag zeige ich, wie Graylog2 funktioniert, wie man seine Logs in das System bekommt und welche technischen und unternehmensrelevanten Nutzen sich daraus generieren lassen. Graylog2 ist ein freies und Open-Source Log Management System, das inzwischen bei Firmen rund um die Welt im Einsatz ist.
Biographie
Lennart Koopmann hat im Jahr 2010 mit der Entwicklung von Graylog2 begonnen und ist Software-Entwickler bei XING.
Nils Ehmke (CAU Kiel): Erstellung und Ausführung von Kieker-Analysen mittels Java-API und Webanwendung, 9:40–10:10 Uhr
Um die schrittweise Verabeitung von Monitoringdaten zu ermöglichen, beinhaltet Kieker eine Java-basierte API zur Definition und Ausführung von Pipe-and-Filter-Architekturen, bestehend aus existierenden oder eigenen Analyse-Plugins. Aufbauend auf dieser API arbeiten wir aktuell an einer Web-basierten Anwendung, die eine graphische Bearbeitung und Ausführung solcher Analyseprojekte erlaubt. Neben der Konfiguration der Pipe-and-Filter-Architekturen mittels eines Editors, lassen sich u.a. sogenannte Dashboards zusammenzustellen, die ausgewählte Ergebnisse der Analyse-Plugins visualisieren. Dieser Vortrag stellt die erwähnte Java-API vor und gibt einen Überblick über den aktuellen Stand und die weiteren Ziele der Webanwendung.
Biographie
Nils Christian Ehmke ist Masterstudent an der Christian-Albrechts-Universität zu Kiel, an der er 2011 zuvor sein Bachelorstudium abgeschlossen hat. Seit 2010 ist Nils als wissenschaftliche bzw. studentische Hilfskraft in der Arbeitsgruppe Software Engineering tätig und entwickelt in diesem Rahmen u.a. am Kieker-Framework mit. Seine Forschungsinteressen umfassen insbesondere Nebenläufigkeit, Parallelität und dynamische Software-Analyse.
Jan Waller (CAU Kiel): Analysis and Visualization of Monitoring Data in 3D, 10:10–10:40 Uhr
Dynamic analysis of software systems with help of the Kieker framework is able to produce huge amounts of data. Goal-oriented monitoring, analysis and visualization are required to focus the view of a systems or performance engineer on important information. This talk will present several different approaches to monitor, analyze and visualize concurrency aspects of an application in 3D. Each approach has a different focus and goal in its execution.
Biographie
Jan Waller is a Ph.D. student and researcher with the Software Engineering Group at Kiel University. He also studied computer science at Kiel University. His research interests include software performance engineering and concurrent systems. Particularly, he is interested in performance testing, monitoring, and evaluation, as well as the visualization of concurrent software systems. Together with a couple of colleagues, Jan is developing the Kieker framework (kieker-monitoring.net).
Matthias Bauer (Consist GmbH, Kiel): Was haben Logdateien mit dem Unternehmenserfolg zu tun? 10:55–11:35 Uhr
Maschinendaten – Logfiles, Performancemonitore, Transaktionslogs, etc. – sind Grundlage der täglichen Arbeit im IT-Betrieb und Anwendungssupport. Daten, ohne die Fehleranalysen nur sehr eingeschränkt möglich sind. Neben den reinen Betriebs- und Supportinformationen enthalten sie vielfach betriebswirtschaftlich relevante Daten, die mit Standardtechnologien heutzutage nur schwer erschlossen werden können. Allein die Bestellung in einem Webshop erzeugt Maschinendaten in diversen heterogenen Systemen und in unterschiedlichen Strukturen und Formaten, z.B. von dem Webshop, dem ERP-System, dem CRM, dem Buchhaltungssystem, der Logistikanwendung und dergleichen mehr.
mehr.
Splunk Inc. – Partner der Consist – ist Hersteller eines gleichnamigen Tools, mit dem die Korrelation und Analyse großer Mengen unterschiedlicher Maschinendaten auf Basis von Big-Data-Technologie derart vereinfacht werden, dass beispielsweise die Mean Time To Repair auf ein Minimum reduziert werden kann. Die Korrelation der Daten ermöglicht darüber hinaus die betriebswirtschaftliche Nutzung der Maschinendaten, z.B. mit Realzeit-Verkaufszahlen für den CEO, Marketinganalysen in Realzeit für den CMO und gleichzeitig Informationen für den ITBetrieb und Anwendungssupport. Alle Informationen stammen aus ohne Transformationsverluste unmittelbar aus ein und denselben Grunddaten.
Biographie
Matthias Bauer ist seit 2006 in verschiedenen Positionen bei der Consist Software Solutions GmbH tätig. Er leitete diverse Produkt- und Projektentwicklungen und ist heute für den technischen Vertrieb sowie die Themen Software-Architektur und eingesetzte Technologien verantwortlich.
Florian Fittkau (CAU Kiel): Online Trace Visualization for System and Program Comprehension in Large Software Landscapes, 11:35–11:45 Uhr
Vortragsform: PechaKucha.
In many companies the number of legacy programs is constantly increasing. Furthermore, those legacy programs often interact with each other forming a software landscape which can include several hundreds of programs. The knowledge of the communication, internals, and utilization of this software landscape often gets lost over the years rendering program and system comprehension more and more important. This talk will propose a solution to support program and system comprehension utilizing online trace visualization. It will present ideas to master the visualization of the huge amount of incoming information in a running distributed software system.
Biographie
Florian Fittkau is a PhD student at the Kiel University and a part-time professional software developer. He received his MSc in computer science from the Kiel University in march 2012. His research interests include program comprehension, distributed systems, empirical methods, and HCI.
Christian Wulf (CAU Kiel): Detection and Utilization of Potential Parallelism in Software Systems, 11:45–11:55 Uhr
Vortragsform: PechaKucha.
Parallel programming is no longer optional to get an increased speed-up. However, efficient parallelization of a sequential program is a challenging and error-prone task. For this reason, we need tools that at least assist the developer and at best fully automatically parallelize the given program.
In order to find and utilize available parallelism potential in software systems, this talk gives an idea of how to use static and dynamic dependence analysis in combination. Furthermore, we propose our pattern-matching solution for both the detection of promising sequential code locations and the generation of parallel versions of them.
Biographie
Christian Wulf is a PhD student at the Kiel University and a part-time professional software developer. He also received his MSc in computer science at the Kiel University. His research interests include software quality as well as parallel and distributed systems. In particular, he is interested in (semi-)automatic techniques regarding the transformation from sequential to parallel code.
