Forschungsthemen
[MA] Towards Latency-aware Self-Optimization of Non-functional Requirements
Typical objectives of self-optimizing software systems are the maximization or minimization of the system's non-functional properties (NFPs). For example, the minimization of cost and the maximization of performance. The basic principle of such approaches is the MAPE-K loop~\cite{KC03}. Within the monitoring phase, the current values of the NFPs are observed. The analyze phase checks whether the NFPs are satisfying the current requirements. If they don't, the plan phase derives a sequence of adaptation actions to be performed against the system to improve the NFPs. Finally, the execution phase actually performs these actions.
This thesis shall investigate a novel approach for self-optimization, which takes the latency of adaptation actions into account. Examples have been created by Javier Cámara and Gabriel Moreno\cite{camara2016analyzing,moreno2018flexible}. This includes in particular that the latency of adaptation actions can be part of the objective function. In other words, the system to be developed cannot just optimize particular NFPs, but also the time required to change them.
The Latency-aware RDM Simulator~\cite{GSB24} shall be used as a case study. As a possible base approach, which is not yet latency-aware, the RE-STORM approach~\cite{GSB24b} can be used. Alternatively, another base approach has to be used and extended.
Betreuer: Sebastian Götz