Open-source code tracks data’s international travels
In a world first, University of Victoria scientists show that labelling data doesn’t slow ultra-high-speed delivery
University of Victoria
image:
Booth at International SuperComputing Conference in St. Louis, Missouri, November 2025
view moreCredit: Bruno Hoeft, KIT, Germany
Imagine packing up a priceless work of art and sending it to a gallery on the other side of the world. Imagine doing that with no paperwork for the courier company. Now imagine that instead of art this is data about subatomic particles, the molecular structures of potentially toxic street drugs, global climate models or the far reaches of the universe.
As detectors and sensors and instruments of all kinds become increasingly sensitive, research generates orders of magnitude more data than ever. Scientists have developed methods and infrastructure to transfer those vast quantities of data more quickly, says Randall Sobie, research scientist at the Institute of Particle Physics. Now, UVic researchers have labelled the data and—critically—proved that the labels did not slow down the process.
Game-changing code
Tristan Sullivan, a high-energy physicist at UVic, developed the open-source code that puts markers on the data. Then he used a technique known as SciTags to identify who owns the data and its purpose. The information could be used to distinguish real data from simulated data, or network tests from data transfers, for example.
The proof came in November 2025 when scientists at UVic sent labelled data at a rate of 1.15 terabits per second to the International SuperComputing Conference in St. Louis, Missouri. That’s like streaming more than 66,000 high-definition movies at once.
This could be a game changer for individual research projects as well as networks like BCNET, which provides high-speed regional connectivity to the research community, and CANARIE, the national ultra-high-speed network that connects researchers in Canada to networks around the globe.
“The packet marking being able to keep up with the data moving at 1.15 terabits per second was an important part of the demonstration," says Tristan Sullivan, high-energy physicist at the Victoria Subatomic Physics & Accelerator Research Centre. "It opens the door to real-time visibility, measurement and optimization of large-scale scientific data movement.”
That matters to individual researchers and projects, and to partners.
Networks and partnerships
“We have Quality of Service agreements with international projects like those at CERN [the European Organization for Nuclear Research in Geneva, Switzerland] and the Square Kilometre Array [of radio telescopes],” explains Ryan Enge, director of Research Computing Services at UVic. “If we guarantee some minimum bandwidth for those projects, packet marking will let us monitor it and make sure each partner is receiving its share.”
Sullivan, Sobie and Enge all emphasize the vital role of the networks that carry the data across countries, over borders and beneath oceans, and industry partners such as Dell, NVIDIA and Lenovo, who supply hardware that makes the volume and speed possible.
“The markers,” Sobie says, “can help us manage our data, monitor it and make better use of networks and storage.”
This new proof that labels won’t slow the process shifts research into a new gear. When scientists receive galactic amounts of data, they’ll know whose it is, why it was collected and what might be done with it. And they’ll know it in less than the time it takes to say “the art of science.”
