DCBX configuration for RoCE

RDMA over Converged Ethernet (RoCE) is a network protocol that enables remote direct memory access (RDMA) over an Ethernet network. There are two RoCE versions: RoCE v1 and RoCE v2. RoCE v1 is an Ethernet link layer protocol that enables communication between any two hosts in the same Ethernet broadcast domain. RoCE v2 is an Internet layer protocol that routes the packets.

HPE Synergy supports lossless RoCE v1 and RoCE v2 traffic types using the Data Center Bridging Exchange (DCBX) protocol. The HPE Synergy interconnect modules identify and treat the RDMA traffic as lossless by implementing DCBX between peer devices. Peer devices can be either of the following:

At Layer 2, the losslessness is achieved by configuring Priority Flow Control (PFC) in DCBX for RoCE traffic. DCBX is always configured between two peers by protocol negotiation (IEEE or CEE).

The DCBX status is displayed on the ports, which are configured for RoCE traffic. You can configure the RoCE traffic class under the QoS and DCBX configuration section. Currently, one traffic class can be assigned to RoCE.

For the RoCE and FCoE combined config type, the aggregated switch represents coexistence of RoCE and FCoE either on the same port or on different ports. The ports can be physical uplink, and downlink ports.

The RoCE configurations are supported on the following devices:
  • HPE Virtual Connect SE 40Gb F8 Module for HPE Synergy

  • HPE Virtual Connect SE 100Gb F32 Module for HPE Synergy

RoCE considerations
  • RoCE requires a separate lossless queue.

  • All RoCE traffic must be tagged Ethernet.

  • RoCE traffic can cross the stacking links on supported devices.

Limitations

  • There is no support for routable RoCE (RoCE v2).

More information