Data Driven Scalability and Cost Analysis for Evolutionary Architecture In the Cloud - Ian Gorton

Otras charlas de Global Software Architecture Summit 2019 también en podcast: https://lk.autentia.com/GSAS19-Podcast An advantage of evolutionary architectures is their facilitation of experimentation and data-driven development. This becomes especially important for cloud-deployed applications as they scale to meet greater loads, while continually balancing costs. Commercial clouds offer a myriad of powerful, competing services that can facilitate scalability through incremental change. However, the performance and scalability characteristics of these services are a priori unknown. They require careful evaluation and testing to ensure they deliver the required quality of service within application budgets. In this talk, we’ll describe the results from a set of experiments on cloud-based micro-benchmarks that explicitly compare cloud services behavior under experimental conditions. The results show how variability in microservice programming language, database platform, database model and runtime configuration settings dramatically effect scalability and associated costs. The results from these experiments can help architects reason about the effects of their architectural decisions and cloud services selections and configurations. We’ll conclude by discussing the implications of this work for evolutionary architecture, and our progress on publishing these results online for architects to access and utilize.