The Anatomy of Networking in High Frequency Trading


PJ Waskiewicz



Session Type




Networking has always served a number of very diverse environments. From Enterprise to the Cloud, Telco and edge, networking technologies have been able to use a “some sizes fit most” approach. This is good when it comes to supporting these technologies in the Linux kernel.

More specialized environments, such as High Frequency Trading (HFT), have radically different networking requirements. Depending on the use case, one requirement might be that latency is paramount when interfacing with the market exchanges. Another use case might be in the HPC environment, where latency is still paramount, but sustained and reliable throughput is a must across grid networks.

This talk is intended to highlight where the Linux kernel networking stack intersects these requirements for HFT, and where it does not. It will also expand on how latency and jitter within HFT systems compare to “traditional” networking environments. Ultimately this talk is intended to generate discussion where HFT networking needs can help improve the existing intersection points in the kernel, and discuss where further native integration could be achieved.