Teh Linux 6.14 kernel has introduced a critically important update to its RDMA subsystem,with the most notable change being the addition of ROCEv2 protocol support within the ERDMA driver. This advancement marks a pivotal step for Alibaba Cloud, as the ERDMA driver is specifically designed for Alibaba Elastic RDMA Adapters (ERDMA).
ROCEv2, the Internet layer protocol for RDMA over Converged Ethernet (RoCE), is now fully integrated into the ERDMA driver. Alibaba engineers highlighted the importance of this update in their patch series, stating:
“This patch series introduces support for the RoCEv2 protocol into the erdma driver. as the most prevalent RDMA protocol, RoCEv2 is widely used in the production environment. Given the extensive submission of erdma across various scenarios in the Alibaba Cloud, there has arisen a requirement for erdma to support the RoCEv2 protocol. Therefore, we update both the erdma hardware and the erdma driver to accommodate the RoCEv2 protocol.”
This enhancement aligns the ERDMA driver with other industry leaders like the NVIDIA Mellanox driver, which has long supported ROCEv2 for remote direct memory access over Ethernet networks. However, it’s worth noting that the ERDMA driver is primarily tailored for alibaba Cloud environments, limiting its utility outside this ecosystem.
The RDMA updates for Linux 6.14 also include a range of other improvements, though none as impactful as the ROCEv2 integration. For a detailed look at all the changes, you can explore the full list of updates via this pull request.
Below is a summary of the key updates in the Linux 6.14 kernel RDMA subsystem:
| Feature | Description |
|—————————|———————————————————————————|
| ROCEv2 Support | added to the ERDMA driver for Alibaba Elastic RDMA Adapters. |
| ERDMA Driver Updates | enhanced to accommodate ROCEv2 protocol, aligning with Alibaba Cloud needs. |
| RDMA Subsystem Changes | Light set of updates, with ROCEv2 being the most significant addition. |
This update underscores Alibaba Cloud’s commitment to optimizing its infrastructure for high-performance computing and networking. While the ERDMA driver may not have broad applicability outside Alibaba Cloud, its integration of ROCEv2 is a testament to the company’s focus on leveraging cutting-edge technologies to meet the demands of modern cloud environments.
For those interested in the technical specifics, the complete RDMA updates can be found in the Linux kernel mailing list.