Network Automation Python-based Application: The performance of a Multi-Layer Cloud Based Solution

Abstract

In recent years, we have witnessed a growing interest in adopting network automation solutions to maximize network availability in increasingly hybrid data centers networks. Amongst all the solution design characteristics, reliability, performance, scalability, and minimal resource overhead, are essential features that influence the adoption choice. This interest, and the associated challenges, have led to a rapid development of several network automation solutions that are available in the market, mostly offered by the original equipment manufacturers (OEMs). In this paper, we explore the ability to develop lightweight network automation Python programs to automate data center network tasks that can run on private cloud environments. We compare the programs performance running on the cloud implementation with a conventional deployment on a standalone physical server. Further, we evaluate the execution time performance and the recovery of the Python programs running different network tasks. Our measurements show that the reading performance for the programs running on the cloud achieves a steady performance across multiple runs with relative improvements compared to the standalone servers. Further, the average performance for the writing tasks produced comparable results for both scenarios. Our analysis shows that the worst writing performance when running on the cloud can still achieve a 22% better performance. Finally, our experimental results show effective utilization of the cloud built-in high availability features to provide necessary levels of recovery without losing the running state, and with acceptable resource overhead.