Derm: SLA-aware Resource Management for Highly Dynamic Microservices

We present Derm, a SLA-aware Resource Management for Highly Dynamic Microservices. To underscore the efficacy of Derm, we have made our artifact available on Zenodo, comprising the complete source code for our proposed implementation. This encompasses all components, such as the scheduler, and functionalities discussed in the paper, including the latency profiler, dynamic predictor, and scale optimizer. Additionally, users have access to example training and profiling scripts, which can be easily adapted to their own models and datasets.  Example configuration and explanation can be found in our codes.