Feasibility Assessment of a Distributed Environment Based on Ray

The Goal 🥅

The goal of the project is the creation of a distributed computing environment based on Python and the Ray framework, designed to unify the computational resources of the three workstations of the Lab. This setup ensures efficient management of parallel tasks and dynamic scalability according to workload demands.

Part 1 – Feasibility Assessment of the Design

The Laboratory is equipped with three high-performance workstations, each featuring an NVIDIA RTX 3080 GPU and an Intel i9 processor, designed to deliver exceptional computing power for simulation, graphic processing, and advanced scientific computation.
These systems form an ideal technological foundation for the creation of a high-performance distributed environment, capable of efficiently managing complex and parallel workloads.

The study and design of the prototype are based on a state-of-the-art open-source solution: Ray, a framework that enables the efficient management of distributed and parallel computation.
The real strength of Ray lies in its ability to create a scalable and flexible computing environment, adaptable to different infrastructures—from a single laptop to complex on-premise systems—and compatible with both Windows and Linux.

This versatility makes it suitable for both small-scale contexts and more complex infrastructures, ensuring high performance and operational reliability.
The first step of the prototype involves the use of Ubuntu on Hyper-V, a configuration that allows the efficiency of Linux to be leveraged within Windows machines, combining the power of the open-source operating system with the compatibility of the Microsoft environment.

Subsequently, Ray is installed within a Kubernetes environment, a platform that enables dynamic scalability and automatic orchestration of resources across multiple nodes.

Thanks to Kubernetes, the prototype can optimally manage distributed workloads, dynamically allocating resources and ensuring stable and high-performance operation.

This configuration represents a strategic step toward building an efficient and modular distributed infrastructure, capable of supporting advanced research and experimental development activities.

Part 2 – Conclusion of the Feasibility Evaluation of the Prototype