After processing on the FPGA, the incoming packets are VLAN-tagged, contain a timestamped Ethernet trailer and are combined into up to 4 100G streams. There are two possible device configuration options for further processing of this traffic.
NeoTap Standard: data processing and routing on the device through backplane
The Standard version of NeoTap device includes one or two 100G NICs and a powerful CPU to receive the data from the FPGA, transform and store it on the device, or forward to its final destination.
- 16-core 3.7 GHz AMD EPYC Milan CPU for standard configuration
- Up to 256 GB RAM for ultra-deep packet buffers
- One or two PCIe 4.0 x16 100G NICs for up to 400 Gbps backplane
- Multiple NIC options: Intel E810, Mellanox Connect-X6, etc.
- Simple VLAN interface mode: work with front panel ports using standard Linux tools: ping, tcpdump, etc.
- DPDK-based high-performance packet routing utility from LDA: receive data from front panel ports, aggregate and rebroadcast through another ports. Source code available for custom modifications: compression, packet manipulation, etc.
- Third-party tools and libraries supported out of the box: all standard Linux and NIC drivers used
NeoTap Basic: no processing on the device itself – data delivered to an external capture system
This is the most cost-effective option for the customers who already invested in a good packet capture and processing infrastructure.
- Up to 4x 100G streams delivered to external infrastructure directly from the FPGA
- No further processing on the NeoTap device – basic, cost-efficient CPU for configuration only
- No ability to rebroadcast through front panel ports
- Currently, only 100G is supported for the FPGA's output data streams