Park, SungjaeThomas, R. QuinnCarey, Cayelan C.Delany, Austin D.Ku, Yun-JungLofton, Mary E.Figueiredo, Renato J.2024-12-202024-12-202024-09979-8-3503-6562-72325-372Xhttps://hdl.handle.net/10919/123861Modern Function-as-a-Service (FaaS) cloud platforms offer great potential for supporting event-driven scientific workflows. Nonetheless, there remain barriers to adoption by the scientific community in domains such as environmental sciences, where R is the focal language used for the development of applications and where users are typically not well-versed with FaaS APIs. This paper describes the design and implementation of FaaSr, a novel middleware system that supports event-driven scientific workflows in R. A key novelty in FaaSr is the ability to deploy workflows across FaaS providers without the need for any managed servers for coordination. With FaaSr: 1) functions are written in R; 2) the runtime environments for their execution are customizable containers; 3) functions access data in cloud storage (S3) with a familiar file-based abstraction supporting both full file put/get primitives and subsetting using the Parquet format; and 4) function invocation and workflow coordination only requires S3 cloud object storage, without relying on any dedicated, active workflow engine server or cloud-specific queues/databases. The paper reports on the functionality and performance of FaaSr for micro-benchmarks and two case studies: event-driven forecast and batch job workflows. These demonstrate the ability to deploy workflows across multiple platforms (GitHub Actions, Amazon Web Services Lambda, and the open-source OpenWhisk), without the need for dedicated coordination servers, across both cloud and edge resources. FaaSr is open-source and available as a CRAN package.10 page(s)application/pdfenIn CopyrightcloudcyberinfrastructureFunction-as-a-ServiceserverlessworkflowLAKEConference proceedinghttps://doi.org/10.1109/e-Science62913.2024.10678660Thomas, Robert [0000-0003-1282-7825]