FMI: Fast and Cheap Message Passing for Serverless Functions

TL;DR

FMI introduces a high-performance, cost-effective message passing framework for serverless functions, enabling efficient distributed computing with significant speedups and cost reductions.

Contribution

This paper presents FMI, a novel message passing interface for serverless functions that supports various communication channels and models MPI, improving performance and cost-efficiency.

Findings

FMI speeds up distributed ML FaaS applications by up to 162x.

FMI reduces communication costs by up to 397x.

Supports portable, MPI-like communication in serverless environments.

Abstract

Serverless functions provide elastic scaling and a fine-grained billing model, making Function-as-a-Service (FaaS) an attractive programming model. However, for distributed jobs that benefit from large-scale and dynamic parallelism, the lack of fast and cheap communication is a major limitation. We overcome this limitation and present the FaaS Message Interface (FMI). FMI is an easy-to-use, high-performance framework for general-purpose point-to-point and collective communication in FaaS applications. We support different communication channels and offer a model-driven channel selection according to performance and cost expectations. We model the interface after MPI and show that message passing can be integrated into serverless applications with minor changes, providing portable communication closer to that offered by high-performance systems. In our experiments, FMI can speed up communication for a distributed machine learning FaaS application by up to 162x, while simultaneously reducing cost by up to 397 times.