TNO – DynAA
| ID: R16 | Licence: Apache 2.0 | Owner: TNO | Contacts: julio.deoliveirafilho@tno.nl |
| Short Description | DynAA is a highly efficient system simulation engine based on discrete event specifications. In MYRTUS it will be used by SAM (possibly FREVO) and will be made compatible with TOSCA system descriptions. |
| Key features | – Fast simulation of combined SW execution profiles – System-wide description of the continuum includes resources at all levels – Able to simulate interconnection between nodes and the effects of data exchanges – Able to serve as a digital twin of the continuum deployment, enabling analysis, estimation, and optimization. – Simulations can be parallelized in HPC or cloud for accelerated evaluations |
| Require | N/A |
| Provide | – (Custom) Detailed simulation traces of system and compound SW component behaviors – Estimation of system KPIs, extracted from trace analysis |
| Input | – Characterization of SW component resource utilization and behavior in advance. – Description of current status and configuration of the system |
| Output | – (Custom) Detailed simulation traces of system and compound SW component behaviors – Estimation of system KPIs, extracted from trace analysis |
| User | In the MYRTUS context, DynAA will be mainly used in combination with the tool SAM on the WL manager. SAM is a DOCP optimization engine that uses DynAA to evaluate system KPIs of possible solutions during space exploration. Eventually, DynAA may be combined in the DPE with FREVO in the same role. |
| Benefits for the user | Quantitative analysis reasoning and optimization for the WL orchestration process. |
| Position in the MYRTUS DPE | Step 1 – Continuum modeling, simulation and analysis |
| TRL@M0 | 5-6 |
| TRL@M36 | 5-6 |
General description
DynAA discrete event engine is fully developed in TNO from other projects and as such a mature and already existing SW module. The engine is written in C++, highly performant and memory efficient — that is, it is written to have a minimal footprint for the simulation models (on memory and performance). DynAA offers a Python integration wrapper — called pyDynAA that eases the integration with prototypes and simulations driven from Python.
Role in the MYRTUS DPE
In the MYRTUS context, DynAA will be mainly used in combination with the tool SAM on the WL manager. SAM is a DOCP optimization engine that uses DynAA to evaluate system KPIs of possible solutions during space exploration. Eventually, DynAA may be combined in the DPE with FREVO in the same role, that is, provider of KPI estimation and analysis for WL managers and orchestrators.
MYRTUS Extension/Contribution
DynAA will be extended with a library of models that translate to basic system descriptions, such as the ones found in TOSCA. This library will facilitate (and partially automate) the simulation of system compositions. Interfaces with MIRTO managers such as the ones based on SAM and FREVO will be implemented with this library. Such simulations can be used by orchestrators and managers at the MIRTO engine.
Plans and Expectation
Assessment Plan@M18:
Initial simulations of system models using DynAAs will be demonstrated. Possibly already in combination with the FREVO tool, which delivers the topology model to be simulated. Assessment made at laboratory.
Expected Results@M18:
Transformation of topology system descriptions into DynAA models (possibly not yet automated) and demonstration of system simulations using DynAA.
References
- TBD