MPN: 90Y3452·By: IBM
Ships from the US.
We'll confirm availability and pricing for 90Y3452, usually within one business day.
We'll confirm availability and pricing for 90Y3452, usually within one business day.
The IBM Flex System IB6131 InfiniBand Switch (SKU 90Y3452) is a managed plug-in module for Flex System chassis, providing 36 InfiniBand FDR/QDR ports including 14 internal ports at 40 Gbps QDR (upgradable to 56 Gbps FDR) and 18 external QSFP uplink ports supporting 10/20/40/56 Gbps auto-negotiation. It offers 2 Tbps switching capacity at FDR and 1.44 Tbps at QDR with full bisectional bandwidth and sub-170 ns node-to-node latency at FDR speeds. Designed for high-performance computing (HPC), enterprise datacenters, and financial services, it supports clustered databases, parallel processing, and transactional services via low-latency fabric. Features include Forward Error Correction, nine virtual lanes for QoS, 48K unicast/16K multicast addresses, and management via 1GbE/RS-232 ports. Compatible with IB6132 mezzanine cards, it reduces CPU overhead through RDMA, enabling efficient scaling for up to 14 servers with zero oversubscription.
| Manufacturer part # | 90Y3452 |
|---|---|
| Brand | IBM |
| Condition | Refurbished |
| Model | IB6131 |
| Standards | IBTA 1.3 and 1.21 compliant |
| Product Line | Flex System |
| Product Type | InfiniBand Network Switch |
| Management OS | PowerPC based MLNX-OS |
| InfiniBand Speeds | 10 Gbps, 20 Gbps, 40 Gbps QDR, 56 Gbps FDR (with upgrade) |
|---|---|
| Unicast Addresses |
No specific compatibility data found in available sources
Supports parallel processing and clustered databases with 2 Tbps FDR switching capacity and sub-170 ns latency, enabling faster task completion for compute-intensive workloads.
Provides high-bandwidth, low-latency fabric for 14 servers via 36 FDR/QDR ports with zero oversubscription, ideal for storage clustering and embedded I/O applications.
Delivers transactional services performance through full bisectional bandwidth and RDMA offload, reducing cost per operation in InfiniBand infrastructures.
Scales to larger node counts with 18 QSFP uplinks and low packet hops, supporting up to 48K unicast addresses for efficient multi-server environments.
Details
Shipping, warranty & returns
No reviews yet — be the first to review this part.
Secure SSL checkout. Visa, Mastercard, Amex and Discover accepted. ACH / wire available for businesses.
New, certified-refreshed and OEM stock with guaranteed compatibility, backed by warranty on every part.
We process quotes, POs, stock documentation and compliance invoicing.
[email protected]| 48,000 maximum per subnet |
| Switching Capacity | 2 Tbps for FDR and 1.44 Tbps for QDR |
|---|
| Multicast Addresses | 16,000 per subnet |
|---|
| Form Factor | Flex System Plug-in Module |
|---|
| Port Type | Managed Plug-in Module |
|---|---|
| Total Ports | 32 ports |
| Console Port | 1 mini-USB RS-232 |
| Network Ports | 36 x InfiniBand FDR/QDR Ports |
| External Ports | 18 QSFP ports |
| Management Port | 1 GBE port |
| Network Interfaces | InfiniBand FDR/QDR; High-speed data transfer |
The switch supports 40 Gbps QDR InfiniBand natively on 36 ports (14 internal, 18 QSFP external) and upgrades to 56 Gbps FDR via Feature-on-Demand license. QSFP ports auto-negotiate 10/20/40/56 Gbps, enabling high-speed uplinks with copper DAC cables for HPC fabrics.
It provides 2 Tbps capacity at FDR and 1.44 Tbps at QDR with full bisectional bandwidth. Node-to-node latency is under 170 ns at FDR speeds—nearly half of QDR—plus Forward Error Correction for resilient, low-packet-loss performance in clustered environments.
Yes, designed as a plug-in module for Flex System Enterprise chassis, connecting 14 internal ports to compute nodes. Pairs with IB6132 dual-port mezzanine cards for FDR/QDR links, integrating into high-bandwidth, low-latency datacenter fabrics.
Includes 1GbE port to chassis management module, RJ-45 Ethernet, and RS-232 mini-USB serial ports. Features PowerPC MLNX-OS with in-band agents; requires external subnet manager. Supports 48K unicast/16K multicast addresses and nine QoS virtual lanes.
Optimized for HPC clustering, parallel processing, transactional services, and embedded I/O in datacenters/financial services. RDMA offload reduces CPU overhead; scales for larger node counts with 18 uplinks and zero oversubscription for 14 servers.
Yes, base QDR (40 Gbps) upgrades to FDR (56 Gbps) via FoD license, requiring compatible IB6131 FDR adapters. Delivers ultra-high performance with low power consumption and IBTA 1.3/1.21 compliance for enterprise-grade reliability.
You might also want to check out these products.

Cisco Meraki MS125-48LP-HW

Cisco Meraki MS125-48LP

AT-8624T/2M - Allied Telesis 8600 24-Port 2x Module Bays uplink Switch

AT-9000/24 - Allied Telesis 9000/52-10 24-Port Switch

469-3414 - Dell PowerConnect 5524 24-Port + 2 x SFP+ Ports 1U Switch

AT-GS910/24-50 - Allied Telesis CentreCOM GS910 24-Ports Switch

AT-FS717FC/SC-10 - Allied Telesis 16x 1x 100BFX SC Connector Switch

AT-9424T/SP - Allied Telesis 9400 24-Port 2x Port SFP Switch