Key technical areas covered in this session include:

• • The AF_XDP Advantage: How bypassing the traditional Linux kernel networking stack eliminates memory-copy overhead, enabling synthetic throughput of up to 98 GB/s for large payloads.
  • Hardware and Tuning: A breakdown of the strict prerequisites for these speeds, including AF_XDP-capable NICs, and the critical role of hugepages and IRQ affinity tuning to minimize CPU jitter.
  • The afxdp Rust Crate: An architectural look at the crate responsible for low-level ring integration and providing zero-copy views for the forwarding pipeline.
  • Safety vs. Raw Throughput: A comparative analysis showing how Ryan maintains Rust-enforced memory safety and Lean 4 formal verification while remaining competitive with industry standards like DPDK and WireGuard