Performance Measurements of Function as a Service Platforms

Abstract

This paper evaluates the current state of the Function as a Service (FaaS) landscape and investigates the extent of the applicability of this new technology for the use-cases of today. We have selected the most popular Function as a Service platforms and have measured the relevant application performance parameters (latency and database access rate) under different types of load. Our source code used by our experiments is available from our public repository. Our measurements confirm that the investigated FaaS technologies have stable performance characteristics under various load conditions. They can be safely applied to a large number of use-cases, and their performance will hold up to the traditional solutions with also added benefits as well, such as less operational costs and better scalability.