Understanding Open Source Serverless Platforms: Design Considerations and Performance

TL;DR

This paper investigates how design choices impact the performance of open-source serverless platforms, revealing that auto-scaling strategies need to consider multiple resource and workload factors for optimal efficiency.

Contribution

It provides an analysis of performance-influencing design issues in open-source serverless platforms and highlights the limitations of simple auto-scaling approaches.

Findings

Performance varies significantly due to design idiosyncrasies.

Single-resource auto-scaling is insufficient for optimal performance.

Multiple factors must be considered for effective auto-scaling.

Abstract

Serverless computing is increasingly popular because of the promise of lower cost and the convenience it provides to users who do not need to focus on server management. This has resulted in the availability of a number of proprietary and open-source serverless solutions. We seek to understand how the performance of serverless computing depends on a number of design issues using several popular open-source serverless platforms. We identify the idiosyncrasies affecting performance (throughput and latency) for different open-source serverless platforms. Further, we observe that just having either resource-based (CPU and memory) or workload-based (request per second (RPS) or concurrent requests) auto-scaling is inadequate to address the needs of the serverless platforms.