Sieťová DPU NVIDIA BlueField-3 B3210E E-Series FHHL, Dual-Port 100GbE QSFP112, 32GB DDR5

The NVIDIA BlueField-3 B3210E E-Series FHHL Network DPU is a data processing unit built for high-performance network offload, security acceleration, and infrastructure services in servers and edge platforms. This FHHL (Full-Height Half-Length) module integrates dual 100GbE QSFP112 ports and 32GB DDR5 memory to handle networking, storage, and security workloads while reducing host CPU utilization and improving overall application throughput.

Advantages

This DPU provides hardware-accelerated packet processing, programmable offload for networking and storage functions, and integrated memory capacity for low-latency operations. With dual 100GbE ports it supports high-throughput deployments and the FHHL form factor enables compatibility with a wide range of server platforms. By offloading complex tasks from the CPU, it helps improve application performance and optimize server resource usage in data center and edge environments.

Suitable for

• High-bandwidth network virtualization and software-defined networking deployments
• Storage offload, NVMe-oF target acceleration, and RDMA-enabled workloads
• Security functions such as firewalling, encryption, and DPI in hardware-accelerated environments
• Data center servers and edge platforms that require reduced CPU overhead and increased I/O efficiency

Key specifications

• Form factor: FHHL (Full-Height Half-Length)
• Network interfaces: 2 × 100GbE via QSFP112 connectors
• Memory: 32 GB DDR5 onboard memory
• Primary functions: programmable DPU for network, storage, and security offloads
• Performance focus: high throughput, low latency packet processing and offload acceleration

How to use

Install the FHHL module into a compatible PCIe slot on the server according to the server vendor's hardware installation guidelines. Connect 100GbE links using compatible QSFP112 transceivers or direct-attach cabling and configure the DPU through the vendor-supplied management and software tools. Deploy supported offload profiles and services—network, storage, or security—via the DPU’s management interface or orchestration platform to shift workloads from the host CPU to the DPU.

Recommendations

For optimal results, verify server compatibility and available PCIe lanes, ensure proper cooling and airflow in the chassis, and use validated QSFP112 transceivers or cables. Update firmware and drivers to the latest vendor-recommended versions and follow vendor guidance when enabling specific offload features to maintain stability and performance.