QuickSpecs
HPE HDR InfiniBand Switches and Cables QuickSpecs
Table of Contents
Mellanox InfiniBand (IB) HDR Switches are supported to work with HPE InfiniBand Adapters to deliver high performance connectivity for HPC clusters using HPE ProLiant XL and DL Server families
The 1U switches feature 40 QSFP56 ports delivering up to 16Tbps of non-blocking bandwidth with sub-90ns port-to-port latency.
Ideally suited for Customers who deploy high-performance computing (HPC) clusters based on HPE ProLiant XL and DL Servers using InfiniBand technology.
Overview
Key features
- – 40 ports, each providing HDR (200 Gbps) performance
- – 80 equivalent ports, each providing HDR100 (100 Gbps) performance, using splitter cables
- – 16Tbps non-blocking aggregate bandwidth with sub-90ns port-to-port latency
- – In-network computing through the Co-Design Scalable Hierarchical Aggregation Protocol (SHArP) technology
- – Quick and easy setup and management
- – Multiple Virtual Lanes (VLs) per physical port
- – Improved performance by removing fabric congestion
- – Backward compatibility to Mellanox FDR and EDR InfiniBand technology for ease of migration
Standard Features
Features and benefits
Switch Specifications
- – Mellanox HDR IB 40P switches are based on Mellanox Quantum™ Switch Silicon 40-port ASIC with SHArP support
- – HDR (200 Gbps) per port bidirectional
- – Switching capacity 16Tbps
- – Virtual lanes, configurable from one to eight VLs plus one management VL
- – Configurable MTU size of up to 4 KB
- – Maximum multicast table size: 16384 entries
- – Support QSFP56 passible copper and active optical cables
- – Compliant with IBTA 1.21 and 1.3
- – Adaptive Routing
- – Port Mirroring
HDR100
The Mellanox Quantum™ switch, when associated with ConnectX®-6 adapters, supports the HDR100 protocol. Two HDR100 adapters attached to an HDR splitter cable can be connected to a single port on the switch. Each adapter can then deliver 100 Gbps of bandwidth. In effect, this turns the Mellanox Quantum™ switch into a 80-port 100 Gbps InfiniBand switch, reducing the equipment required for a given fabric when compared to the previous generation and expanding the maximum number of nodes of a 100 Gbps two-level 1:1 fat tree topology to 1,600.
Management
- – Unmanaged switches
- Require external subnet manager on the fabric (see more description below)
- – Managed switches
- Embedded subnet manager supporting fabrics of up to 2K nodes
- Enables command line interface and chassis management GUI through dual 100/1000 Ethernet ports
- RS232 port over DB9
- Enables serial console through USB serial port
- Supports Management over IPv4 or IPv6
- Supports SNMP v1, v2, v3
Connectors and cabling
- – QSFP56 ports
- – Passive copper or active optical cables
Indicators
- – Per QSFP56 port status LED for link and activity
- – System status LEDs for system, fans, power supplies
- – Unit ID (beacon) LED
Power Supply
- – Dual redundant (1+1) hot-plug power supplies
- – Frequency: 50-60 Hz , single phase AC
Cooling
- – Front-to-rear (power side inlet) airflow
- – 6 (N+1) Hot-swappable fan units
Warranty and support
- – Switches carry a 3-year warranty, On Site, next business day response
- – Cables carry a 1-year warranty, parts exchange
Service and Support
For the most up-to-date information on HPE Services, please refer to the HPE Services – Supplemental QuickSpecs, which provides a comprehensive and regularly updated overview of available services.
Configuration Information
| Description | SKU |
| Mellanox InfiniBand HDR 40-port QSFP56 Managed Back to Front Airflow Switch | P06249-B21 |
| |
Related Options
HPE InfiniBand HDR cables | |
| Passive Copper cables | SKU |
| HPE InfiniBand HDR/Ethernet 200GB QSFP56 to QSFP56 1.5m Direct Attach Copper Cable | P06149-B23 |
| HPE InfiniBand HDR/Ethernet 200GB QSFP56 to QSFP56 2m Direct Attach Copper Cable | P06149-B24 |
| Active Optical cables | |
| HPE InfiniBand HDR/Ethernet 200GB QSFP56 to QSFP56 3m Active Optical Cable | P06153-B21 |
Technical Specifications
Mellanox InfiniBand HDR 40-port QSFP56 Managed Back to Front Airflow Switch | P06249-B21 | ||
|---|---|---|---|
| I/O ports and slots | 40 QSFP56 ports | ||
| Additional ports and slots | Dual 100/1000 Ethernet ports 1 RS232 port over DB9 1 USB port | ||
| Power supplies | 2 power supplies (1+1 redundant and hot-swappable) | ||
| Fan units | 4 fan units (N+1 redundant and hot-swappable) | ||
| Physical characteristics | Dimensions | 1.7 in (H) x 17 in (W) x 23.2 in (D) 4.36 cm (H) x 43.32 cm (W) x 59.06 cm (D) | |
| Weight | 12.49 kg 27.536lb | ||
| Management processor | Dual-core x86 CPU for the onboard subnet manager | ||
| Performance | Latency | Sub-90ns port-to-port | |
| Throughput | up to 16 Tbps | ||
| Linear forwarding data base | 4x48K entries | ||
| Environment | Operating temperature | 32°F to 104°F (0°C to 40°C) | |
| Non-Operating temperature | -40°C to 70°C | ||
| Operating humidity | 10% to 85% | ||
| Operating Altitude | Up to 3050 m | ||
| Acoustic | ISO 7779 ETS 300 753 | ||
| Electrical characteristics | Frequency | 50/60 Hz | |
| Voltage | 100-127VAC, 200-240VAC | ||
| Typical power with passive cables (ATIS) | (TBD)W | ||
| Max power with optical cables (assuming 3.5 W per port) | (TBD)W | ||
| Safety | US/Canada: cTUVus EU: EN60950-1 International: IEC60950-1 Customs Union: Russian Federation, Belarus and Kazakhstan | ||
| Emissions | USA: FCC, Class A Canada: ICES, Class A EU: EN55022, Class A EU: EN55024, Class A EU: EN61000-3-2, Class A EU: EN61000-3-3, Class A Japan: VCCI, Class A Australia/New Zealand: AS/NZS CISPR 22, Class A Korea : KCC (KN32, KN35), Class A Taiwan : BSMI (CNS 13438), Class A | ||
| Others | RoHS-6 compliant | ||
| Management | Internally managed switch comes with a dual-core x86 CPU for the onboard subnet manager, enabling simple, out-of-the-box fabric bring-up for up to 2k nodes. The InfiniBand fabric can also be managed by the Mellanox Unified Fabric Manager™. The Mellanox Unified Fabric Manager Advanced (UFM) is a powerful platform for managing scale-out computing environments. UFM enables data center operators to efficiently provision, monitor and operate the modern data center fabric. UFM runs on a server and is used to monitor and analyze Mellanox fabrics health and performance. UFM can also be used to automate provisioning and device management tasks. For example, UFM can communicate with devices, to reset or shut down ports or devices, perform firmware and software upgrades, etc. UFM's extensive API enables it to easily integrate with existing management tools for a unified cluster view. UFM also includes the ability to save historical information, to send alerts to external systems and to activate user made scripts based on system events. | ||
Standards and protocols
IBTA 1.2.1 and 1.3
Summary of Changes
| Date | Version History | Action | Description of Change |
|---|---|---|---|
| 21-Sep-2026 | Changed |
| |
| 16-Feb-2026 | Changed |
| |
| 18-Dec-2023 | Changed | HPE Services Rebranding | |
| 10-Jul-2023 | Changed | Related options section was updated. | |
| 15-Nov-2021 | Changed | Service and Support section was updated. | |
| 05-Aug-2019 | Created | New QuickSpecs |
© Copyright 2026 Hewlett Packard Enterprise Development LP. The information contained herein is subject to change without notice. The only warranties for Hewlett Packard Enterprise products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. Hewlett Packard Enterprise shall not be liable for technical or editorial errors or omissions contained herein.
To learn more, visit http://www.HPE.com/
a00062184enw, - 16384 - Worldwide - V6 - 21-September-2026