Demonstration: Network Traffic Transfers at 200Gbps
CESNET has a long history in the area of high-speed network monitoring using FPGA accelerated cards. Now, we are ready to present our latest FPGA card designed to push the achievable traffic processing throughputs to 200Gbps. The live demonstration will show functionality and unique features of hardware acceleration card with two 100G Ethernet QSFP28 interfaces, powerful Virtex UltraScale+ FPGA, multiple QDR/DDR memories and two PCI gen3 x16 interfaces. Fast DMA engines and highly optimized Linux drivers are used to achieve 200Gbps combined data transfers through both PCIe buses with minimal CPU utilization. Captured network traffic can be distributed among two CPUs (NUMA nodes) and their individual cores based on configurable hash functions. This means that the FPGA card enables wire-speed packet capture to the host memory at 2x100Gbps and can utilize the power of the latest FPGAs and CPUs for various network monitoring applications. Wire-speed packet capture is essential for precise network security monitoring, troubleshooting, and forensic analysis. Support of two ports on a single card is crucial for capturing both directions of traffic (uplink and downlink) on a monitored 100GbE link.
The demo will show how the packets can be received from two fully saturated 100Gbps links and captured to the host memory at 200Gbps without loss. The opposite direction of communication will also be shown, i.e. how the packets can be transmitted from the host memory towards the network interfaces at 200Gbps. Achieved processing speeds will be demonstrated by counters and graphs showing generated, received/transmitted and captured packets. We will also show detailed statistics of CPU load during the packet capture/transmission for various packet lengths.