Tracing Mystery Network Traffic with eBPF Across HRT's Global Compute Cluster

Hudson River Trading (HRT) has been using eBPF (extended Berkeley Packet Filter) to trace and analyze network traffic across its global compute cluster. The company's research team has developed a tool that uses eBPF to capture and analyze network traffic at the Linux kernel level, allowing for detailed insights into network behavior. This is particularly useful for HRT's high-frequency trading operations, where even small delays in network performance can have significant impacts on trading outcomes. The tool has been deployed across HRT's global network, with the company reporting improved network visibility and reduced latency.