FEDS: An Intuitive Model-Driven Middleware for Automated Orchestration and Resource Configuration Across Federated Testbeds
Year: 2025
Type: article
Abstract: As modern and responsive distributed systems grow in complexity, system designers and researchers face challenges in testing these systems at scale under varied workload conditions to evaluate critical attributes such as performance, reliability, security, and maintainability. Although numerous testbed platforms have emerged to facilitate large-scale experimentation, these platforms are designed primarily to study concepts such as virtualization, software-defined networking, and IoT in isolation, thereby leading to fragmented testbeds with limited interoperability. Yet, comprehensive evaluation of modern distributed systems demands validation across multiple testbeds at once, which is hindered due to lack of interoperability, steep learning curve, differing testbed usage policies, and the complexity of resource reservations. Middleware holds the key to overcoming these challenges. In this paper, we introduce the Federated Experiments Design Studio (FEDS), a novel model-driven middleware ecosystem that simplifies the federation of disparate testbeds and automates experiment configuration. FEDS features a graphical front-end with a drag-and-drop interface, allowing users to intuitively design federated topologies, specify experiment properties, and generate the required orchestration artifacts, all within a single tool. This approach automates the entire workflow, from stitching testbeds together to configuring provisioned resources with the necessary packages and system settings. By streamlining the setup process, FEDS enables researchers to focus on rapidly generating valuable experimental insights rather than expending substantial efforts in managing complex infrastructure.
Source: Journal of Systems Architecture
Authors: Sanjana Das, Ruiqing Lan, Caroline Rinks, Yogesh Barve, Abdelilah Essiari, Ezra Kissel, Xi Yang, Paul Ruth, Aniruddha Gokhale
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,540
Funder: National Science Foundation
OpenAlex ID: W4414605216
