From Stateless to Stateful: A Comparative Analysis of Stateful Serverless Computing Frameworks
摘要
The adoption of serverless computing in cloud service architecture has experienced substantial growth, mostly attributed to its inherent scalability and pay-as-you-go payment mechanism. This approach minimizes the need for developer intervention. The initial implementation of Functions-as-a-Service (FaaS) involved stateless operations, however, recent advancements in frameworks have introduced the ability to handle stateful operations. Several stateful serverless systems (SSS) have been developed, such as Apache Flink StateFun (StateFun), Azure Durable Functions (DF), Kalix, and Cloudburst. Nevertheless, the absence of comprehensive comparison criteria poses a significant obstacle to the efficient evaluation and selection of these systems. This article offers a comprehensive examination of different stateful serverless systems, with a detailed analysis of their architectural, functional, and performance characteristics. In addition, we present a new evaluation metric and do a comprehensive analysis of four cutting-edge single sign-on systems. This comparative study provides cloud architects and developers with vital information to make informed decisions when choosing the most appropriate stateful serverless solution for their application requirements.