Abstract: Elasticity is a key property of cloud computing. However, elasticity is offered today at the granularity of virtual machines (VMs) that take tens of seconds to start. This is insufficient to react to load spikes and sudden failures in latency-sensitive applications, leading users to resort to inefficient and expensive overprovisioning. Functions as a Service (FaaS) provides significantly higher elasticity than VMs. However, it is coupled with an event- triggered programming model and a constrained execution environment, making it unsuitable for off-the-shelf cloud applications. Previous work has tried to overcome these obstacles but often requires re-architecting the applications. This paper shows how unmodified off-the-shelf VM-based cloud applications can transparently benefit from ephemeral elasticity using FaaS. We built Rapid, a system that forms an interposition layer spanning VMs and FaaS functions that intercepts execution of unmodified applications and emulates the unified network-of-hosts environment that cloud applications expect when deployed in a conventional VM/container environment. The ephemeral elasticity of Rapid based on AWS EC2 VMs and AWS Lambda FaaS functions enables significant performance and cost savings for off-the-shelf applications with, e.g., recovery times over 5x faster than only using EC2 instances and absorbing load spikes comparable to keeping an overprovisioned pool of EC2 instances
@Article{boxer2024,
title = {Boxer: FaaSt Ephemeral Elasticity for Off-the-Shelf Cloud Applications},
author = {Michael Wawrzoniak, Rodrigo Bruno, Ana Klimovic, Gustavo Alonso},
year = {2024},
eprint = {2407.00832},
archivePrefix = {arXiv},
primaryClass = {cs.DC},
journal = {arXiv:cs.DS:2407.00832}
}