@inproceedings{pavlenko2024vasim,
author = {Pavlenko, Anna and Saur, Karla and Zhu, Yiwen and Kroth, Brian and Cahoon, Joyce and Camacho-Rodríguez, Jesús},
title = {VASIM: Vertical Autoscaling Simulator Toolkit},
booktitle = {IEEE International Conference on Data Engineering (ICDE 2024)},
year = {2024},
month = {May},
abstract = {In recent years, autoscaling has garnered significant attention in cloud computing, emphasizing cost efficiency, performance optimization, and availability for dynamic workloads. New algorithms for horizontal, vertical, and hybrid scaling, targeting instances, VM specifications, and resources like CPU, memory, and IO, have emerged. Various approaches, including forecasting and custom autoscaling functions, are used. However, conducting comprehensive end-to-end testing remains a complex and costly endeavor due to the variety of technology constraints involved.

This paper introduces VASIM, an autoscaling simulator toolkit designed for testing recommendation algorithms, with a particular focus on CPU usage in VMs and Kubernetes pods. The toolkit replicates common components found in autoscaler architectures, including the controller, metrics collector, recommender, and resource updater. It enables a comprehensive simulation of the entire autoscaling system’s behavior, with the flexibility to customize various parameters. In our demonstration, we showcase VASIM’s versatility across multiple use cases, highlighting its effectiveness in evaluating autoscaling strategies, fine-tuning parameters, comparing algorithm performance, and addressing autoscaling-related challenges. This underscores VASIM’s critical role in expediting algorithm development and refinement by providing a controlled environment for testing and experimentation.},
url = {http://approjects.co.za/?big=en-us/research/publication/vasim-vertical-autoscaling-simulator-toolkit/},
}