Formal Foundations of Serverless Computing
share
A robust, large-scale web service can be difficult to engineer. When demand spikes, it must configure new machines and manage load-balancing; when demand falls, it must shut down idle machines to reduce costs; and when a machine crashes, it must quickly work around the failure without losing data. In recent years, serverless computing, a new cloud computing abstraction, has emerged to help address these challenges. In serverless computing, programmers write serverless functions, and the cloud platform transparently manages the operating system, resource allocation, load-balancing, and fault tolerance. In 2014, Amazon Web Services introduced the first serverless platform, AWS Lambda, and similar abstractions are now available on all major clouds. Unfortunately, the serverless computing abstraction exposes several low-level operational details that make it hard for programmers to write and reason about their code. This paper sheds light on this problem by presenting λ_λ, an operational semantics of the essence of serverless computing. Despite being a small core calculus (less than one column), λ_λ models all the low-level details that serverless functions can observe. To show that λ_λ is useful, we present three applications. First, to make it easier for programmers to reason about their code, we present a simplified semantics of serverless execution and precisely characterize when the simplified semantics and λ_λ coincide. Second, we augment λ_λ with a key-value store, which allows us to reason about stateful serverless functions. Third, since a handful of serverless platforms support serverless function composition, we show how to extend λ_λ with a composition language. We have implemented this composition language and show that it outperforms prior work.
READ FULL TEXT