Simplifying Software System Monitoring through Application Discovery with ExplorViz SSP ’18 9th Symposium on Software Performance 2018 Alexander Krause, Christian Zirkelbach, Wilhelm Hasselbring Talk by Sören Henning – Kiel University November 08, 2018
ExplorViz in a Nutshell Live trace visualization of large software landscapes for comprehension of systems and applications [1] Selected Challenges • Finding abstractions to understand huge landscapes but also application- level details • Live visualization of thousands or even millions of traces 2
The ExplorViz Method 3
Let’s start to visualize 4
Let’s start to visualize 5
Let’s start to visualize 6
Let’s start to visualize 7
Let’s start to visualize 8
Let’s start to visualize … with ADAMMS (Application Discovery and Monitoring Management System) 9
Design – Application Discovery OSJPL – Operating System Java Process List 10
Design – Application Discovery OSJPL – Operating System Java Process List 11
Design – Monitoring Management Monitoring ADA MM S Instruction Kill process Update Insert model identifier Internal model Insert Find process Monitoring Get new Run exec OSJPL OSJPL cmd 12
Design – Monitoring Management Monitoring ADA MM S Instruction Kill process Update Insert model identifier Internal model Insert Find process Monitoring Get new Run exec OSJPL OSJPL cmd 13
Design – Monitoring Management Monitoring ADA MM S Instruction Kill process Update Insert model identifier Internal model Insert Find process Monitoring Get new Run exec OSJPL OSJPL cmd 14
Implementation – Software Stack Backend Frontend Agent HTTP (REST) HTTP (REST) 1..* 1..* Cytoscape.js Client Client 15
Implementation – Overview Page 16
Implementation – Process Page 17
First Pilot Study Goals • Overall usability regarding setup and Usability Evaluation operation Setup Results & Discussion 18
First Pilot Study Goals • Overall usability regarding setup and Usability Evaluation operation Experiment Setup • Proband solved tasks • Conductor notated issues • Pluralistic walkthrough [2] Results & Discussion 19
First Pilot Study Goals • Overall usability regarding setup and Usability Evaluation operation Experiment Setup • Proband solved tasks • Conductor notated issues • Pluralistic walkthrough [2] Results & Discussion Structural Interview • Perceived usability • Enhancements 20
First Pilot Study Usability Evaluation Setup Software Landscape • 8 running applications • Tomcat with JPetStore 6 Results & Discussion • kiekerSampleApplication 1 21 1 https://github.com/czirkelbach/kiekerSampleApplication
First Pilot Study Usability Evaluation Setup Software Landscape • 9 running applications • Tomcat with JPetStore 6 Results & Discussion • kiekerSampleApplication 1 22 1 https://github.com/czirkelbach/kiekerSampleApplication
First Pilot Study Learnability • Setup requirements are manageable • Monitoring Management is easily accessible … Usability Evaluation • … but requires knowledge about program internals Efficiency • Faster execution of repeating tasks Setup Memorability • Workflow seemed to be memorable … • … but casual users might show another result Low error rate Results & Discussion • System helped to resolve errors during operation … • … but did not prevent flaws Satisfaction • Graph design is pleasant to use • Management dialogs require more work [3] 23
Conclusions & Future Work Observations indicate good usability More research required • Wider range of probands • Prevent errors (Simple) Monitoring Management works • Provide templates • Refactor GUI forms • Application Discovery method works JDK 9 usage? • Cryptic execution commands? 24
Live Demo 25
References F. Fittkau, A. Krause, and W. Hasselbring. “Software landscape and [1] application visualization for system comprehension with ExplorViz .” In: Information and Software Technology (2016). http://dx.doi.org/10.1016/j.infsof.2016.07.004 J. Nielsen. „Usability Inspection Methods .“ [2] In: Proceedings of the Conference on Human Factors in Computing Systems (CHI ’94) . A. Holzinger. “Usability Engineering Methods for Software Developers.“ [3] Communications of the ACM 48.1 (Jan. 2005), pages 71 – 74 26
Recommend
More recommend