A Petri Net-based Formal Modeling for Microservices Auto-scaling
Souheir Merkouche, Chafia Bouanaka, Elhadj Benkhelifa · 2023
Microservices auto-scaling is an attracking research domain focusing on the dynamic definition of strategies to optimize system efficiency and performance while minimizing service cost. An efficient auto-scaling strikes a balance between individual microservices quality requirements while collaborating to uphold the overall system quality. Maintaining such balance needs a formal method to establish relevant parameters to ensure quality attributes and enable prior accuracy and efficiency verification. To that end, we extend the Petri net model to define Hierarchical Parallel Petri nets (HPPNs) to correctly model both autonomy and cooperation of microservices. Unlike conventional Petri nets, HPPNs consider trade-offs between tokens during transitions, enabling the computation of compromises between multiple quality dimensions and the adaptation strategies. Additionally, preliminary testing mechanisms define the qualities of individual components, providing a solid foundation for assessing microservices’ performance and quality attributes.