Sparking Innovation: How Global Research Networks Enable Discovery

Meeting the Demands of Data-Intensive Science

Modern instruments such as the Large Hadron Collider (LHC) and the Vera Rubin Observatory generate unprecedented quantities of data. For example, the LHC produces tens of petabytes annually, and the Rubin Observatory is expected to generate 20 terabytes of imaging data each night once operational. Transferring and analyzing this data requires networks that are not only fast but also intelligent and adaptive.

That’s where GNA-G comes in. Through its Data Intensive Sciences Working Group, led by physicist Harvey Newman of Caltech, GNA-G is advancing the capabilities of research networks to support real-time data analysis, distributed computing, and autonomous system orchestration.

GNA-G, SENSE, and the National Research Platform

GNA-G’s efforts are deeply intertwined with key network architecture initiatives such as the Software-defined network for End-to-end Networked Science at Exascale (SENSE) and the National Research Platform (NRP). SENSE and NRP provide the programmable, high-performance fabric that connects scientific instruments, computing resources, and researchers.

Recent meetings of the GNA-G Data Intensive Sciences Working Group have showcased how SENSE is being extended with features like asynchronous operation, multi-path routing (ECMP and UCMP), and real-time resource provisioning. These capabilities are critical for enabling automated workflows, such as transferring data from a telescope in Chile to GPU clusters in California or Europe for rapid processing.

From the LHC to the Stars: Collaborative Science in Motion

GNA-G’s work supports collaborations involving major scientific projects. The LHC remains a flagship example, as its global computing grid depends on high-throughput, intelligently managed networks to distribute data from CERN to research centers worldwide. But it’s not just particle physics. The Rubin Observatory, with its nightly sky surveys, relies on robust networking to enable scientists to respond quickly to transient astronomical events.

This infrastructure also supports emerging efforts in radio astronomy, genomics, climate modeling, and AI-driven analysis. With funding from agencies like NSF and DOE, projects are underway to build new testbeds and upgrade backbone connections to 400G and 800G capacities.

Building Smarter Networks

Beyond capacity, intelligence is the next frontier. AI-enhanced routing, telemetry, and decision-making are being prototyped through collaborations with organizations like ESnet and Qualcomm. As part of this effort, GNA-G teams are benchmarking machine learning models for optimizing routing, monitoring transceivers for early failure detection, and using programmable hardware like FPGAs to accelerate data movement.

In one example, researchers are integrating Jupyter AI and lightweight LLMs into the NRP infrastructure to assist with network diagnostics and automation. This work not only enhances operations but also provides a compelling model for future AI-integrated research infrastructure.

At SC25, GNA-G is organizing a series of interactive demonstrations that showcase how international partners are building the networks of the future. Attendees can experience real-time orchestration across transcontinental links, autonomous bandwidth provisioning via SENSE, and AI-assisted workflows supporting science applications. Many of these demonstrations will be taking place at the booths of participating institutions, see right.

These exhibitors, among others, are collaborating with GNA-G to create a unified showcase of global network innovation. A demonstration diagram is in development to help attendees visualize how these efforts interconnect across continents and scientific domains.

KAUST (Booth 4024)

Highlighting international collaboration and real-time data movement across the Middle East and North America.

Starlight (Booth 3131)

Demonstrating programmable networks and distributed data architectures.

SDSC (Booth 217)

Showcasing intelligent workflows and network automation powered by NRP and SENSE.

California Institute of Technology/CACR (Booth 2824)

A leader in developing cutting-edge networks, distributed systems, and innovative approaches for data-intensive science.

Department of Energy (Booth 3802)

A leading provider of high-performance computers, with national laboratories developing and deploying powerful supercomputers to solve scientific challenges.

SC25’s exhibition floor is a vibrant hub of innovation, where attendees can explore cutting-edge technologies, meet with experts, and engage directly with the teams powering next-generation scientific infrastructure.

We invite you to connect with the GNA-G community by visiting their collaborators’ booths, joining live demonstrations, and learning more about how programmable, intelligent networks are enabling global science. Whether you’re curious about AI-integrated routing, real-time data workflows, or how to get your institution involved, this is your chance to see the future in action.

Don’t miss the opportunity to explore the full range of innovation on the show floor, from high-performance networking to groundbreaking visualization and automation tools. SC25 is where science meets scale, and you’re invited to be part of it.