PCM Development/Palladio Coding Session/20180608 SimuLizar Developer Meeting
< PCM Development | Palladio Coding Session(Weitergeleitet von PCM Development/Palladio Coding Session/20180525 Coding Session)
Organizational
- Date: 08.06.2018
- Location: Karlsruhe
Agenda
Simulizar
- Requests
- handle stereotyped elements special (e.g. stereotype to additional behavior)
- UI for measurement point specification
- Language or concept for sensors that are affected by reconfiguration including semantical linking of related sensors
- Reuse background magic that is applied when specifying Simulizar experiment runs in experiment automation and unify specification process
- Create standard measurement points and monitors automatically at simulation start and save measurements in EDP2 properly
- Cleanup
- Extension points review at coding session (many available, partially overlap)
- Registration of new measurement results in runtime model has no perfectly matching measurement point
- Central event handler, InterpreterListener with 25 listeners for all types.
- Extension points review at coding session (many available, partially overlap)
Meeting Notes
- Current features and extensions of SimuLizar (09:30 - 10:30)
- Action model
- Reconfiguration engines
- QVTo: Stable
- Henshin: Essential for non-simulated analysis (model checking)
- StoryDiagrams: Partially stable (unrolling the state space), partially alpha (performance simulation)
- Requests / wishes
- Java / XTend code (simple interface!)
- QVTd
- ATL (??)
- Reconfiguration rules
- Decide between multiple reconfigurations
- Workaround: Have one master QVTo file and delegate
- Non-existent support for elaborate reconfiguration rule management
- There is some code in the master
- Decide between multiple reconfigurations
- Current state of the documentation
- SimuLizar Wiki Page: Core Features w/o Basic Instrumentation (Monitors)
- Simulation Execution missing documentation
- Tutorial from S.Lehrig, pdf file on github: core + instrumentation + ATs + slos + EDP2
- Cloud Scale Book
- Dissertations: M.Becker, S.Lehrig, C.Stier
- QuAL Docu (incomplete)
- Capture current state of the architecture
- Extensibility (-11:50)
- Current state (extension points, etc.)
- Redundancies?
- Simulate stereotyped elements
- Current state (extension points, etc.)
- Lunchbreak (11:55 - 13:30)
- Usability (13:30-14:00)
- Current state and ongoing activities
- Usability improvement for measuring points and monitor repositories
- Automatically generate common measuring types
- Mutliple ways to get to the specifications (graphical editor, wizard, ...)
- Task of the group of students to develop concrete concepts and implement them
- Project running until beginning of October
- 7 Students á 12ECTS
- Perhaps a topic for SSP (depending on the outcome)
- Focus on user experience
- Generate standard measurement points and monitor repositories
- UI for measurement point specification
- In later iterations also experiment automation
- Experiment Automation DSL also @ FZI
- Usability improvement for measuring points and monitor repositories
- Small improvements
- Integrate SimuLizar with the Simulatoin Dock
- Not so small improvements
- Nasty Sirius bugs
- C/P issues
- Unique name constraint
- Compartment handling
- Allocation diagram bug
- Nasty Sirius bugs
- Current state and ongoing activities
- Documentation
- TODO: For each cluster (feature/extension) one page of documentation in the wiki
- TODO: Make S.Lehrigs Tutorial more visible --> wiki
- Pandoc
- Move QuAL documentation to wiki?
- Pandoc ftw.
- Something on Usage Evolution from the Cloud Scale?
- There is nothing on tooling
- Some conceptual stuff, but using CloudScale IDE
- Concerning AT videos from cloud scale could be used
- TODO: Steffen to call CEO from xlab (hosted wiki for CloudScale)
- Documentation
- Design Refactorings (14:00 - 15:30)
- Prototype of separation: Interpreter handler, traversal, event notification
- Interpreter: Behavior of single actions
- Traversal: Decision on next RD behavior (including inner behavior) to process (tree-like)
- Separate stream of interpreter events (simple pub/sub library)
- as declarative as possible
- Prototype of separation: Interpreter handler, traversal, event notification
- Selection of reconfigurations to run
- remove existing work ("reconfigurationrule")
- maybe as conceptual starting point
- topic for another students thesis? (--> TODO Max)
- remove existing work ("reconfigurationrule")
- Selection of reconfigurations to run
- Simulation Models / Simulation Model Registry
- IModelAccess necessary?
- TODO: move Local/Global model copy to ModelCache and onlyuse it in user creation
- Future: Flag: failures on entities removed by reconfiguration
- IModelAccess necessary?
- Simulation Models / Simulation Model Registry
- Language or concept for sensors that are affected by reconfiguration including semantical linking of related sensors
- TODO: Convert broken ResourceURIMeasuringPoints to StringMeasuringPoints on loading of the model
- Conceptual solution: "semantic link", e.g. autoscaling group
- Reference only on group level
- For all aggregated measurements
- Student's thesis? Too complex? Probably not suitable for someone without experience
- Language or concept for sensors that are affected by reconfiguration including semantical linking of related sensors
- Registration of new measurement results in runtime model has no perfectly matching measurement point
- TODO: Check implication of removing all pcm measuring points w/o additoinal information or document necessesity (or find documentation)
- Measuring Points with more than one element reference make sense
- --> Should go into QuAL docu
- TODO: Steffen, Christian, Sebastian dig into potential ways of improving the concept
- Registration of new measurement results in runtime model has no perfectly matching measurement point
- Code quality in general
- TODO: Target top 10 classes w/r/t lines of code
- Code quality in general
- Relation to other tools
- Experiment Automation
- Reuse background magic that is applied when specifying Simulizar experiment runs in experiment automation and unify specification process
- Experiment Automation
- Organizational Stuff
- TODO Documentation (One page per topic, general concept, usage, (optional) development)
- Dealines:
- Initial version: 09.07.18 (next Palladio Concall)
- Version with corrected references: 13.08.18
- Action: Christian
- Power: Christian
- Cloud Metrics: Steffen
- Usage Evolution: Floriment (+ Christian)
- Update Core, include Tutorial from S.Lehrig: Markus F.
- Monitors / MPoints: Students of Markus F.
- Technical page (Transformations): Sebastian
- Overview on extensions: Floriment together with page on ATs
- ATs: Markus F. & Floriment
- Native simulation of events: Dominik
- Arbitrary load balancing strategies: Patrick Firnkes
- Dealines:
- Check if access to wiki for externals is possible: Dominik
- First Prototype of Next Gen SimuLizar: Sebastian, 02.07.18; Review by Steffen and Jörg
- Max: Student's thesis announcement (reconfiguration rules NG) 09.07.18
- Deadlines for the following points: Progress report until next CC, finished by the one after that.
- Remove IModelAccess: Max
- Convert broken ResourceURIMeasuringPoints to StringMeasuringPoints on loading of the EDP2 repository: Markus F.
- Check implication of removing all pcm measuring points w/o additional information or document necessesity (or find documentation): Dominik
- Dig into potential ways of improving the measuring point concept: Steffen, Christian, Sebastian, plan telco during next CC
- Target top 10 classes w/r/t lines of code: Defer to next concall
- Add discussing impact of changes to the release of Palladio to the agenda of the next CC: Stephan
- TODO Documentation (One page per topic, general concept, usage, (optional) development)