Hi All,
We are currently evaluating the network switch requirements for a new Proxmox deployment and would like to understand whether RDMA support is required from the Proxmox or storage perspective.
Our planned environment is:
Any recommendations or best practices would be greatly appreciated.
We are currently evaluating the network switch requirements for a new Proxmox deployment and would like to understand whether RDMA support is required from the Proxmox or storage perspective.
Our planned environment is:
- Proxmox VE
- Local ZFS storage on some clusters
- Third-party shared storage using iSCSI on other clusters
- Standard Kubernetes/VM workloads
- Does Proxmox VE itself make use of RDMA for any of its core features (live migration, clustering, ZFS, etc.)?
- For local ZFS storage, is there any benefit or requirement to use RDMA-enabled NICs?
- For third-party iSCSI storage, does RDMA depend entirely on the storage vendor (iSER/RoCE/iWARP support), or does Proxmox provide any native RDMA integration?
- Is RDMA a supported and commonly deployed configuration with Proxmox VE, or is standard Ethernet networking the recommended approach?
- If RDMA is used with iSCSI or another storage solution, would you recommend selecting switches that support DCB features such as PFC, ETS, and DCBX?
- Has anyone successfully deployed Proxmox with RDMA-enabled networking? If so, which NICs, storage platform, and switch models are you using?
Any recommendations or best practices would be greatly appreciated.