Performance limitation of the MR216i-p controller

[Rob Leadbeater]

We've recently purchased a DL380 Gen11 server, configured with 2 x MR216i-p cards.

It's a 24SFF CTO model, with 4 x NVMe drives and 16 x SAS HDDs. The drives are configured in JBOD mode, as they're due to be used for Storage Spaces.

All three storage enclosures in the server are the 8 SFF 24G x1 NVMe/SAS model, P20882-001, with the single port on each enclosure connected to one of the ports on the RAID cards.

Am I right in thinking that this configuration will be limiting the speeds of the NVMe drives?

From the specs, the drives (P65205-001) should be capable of 3676 MiB/s write throughput, but I'm measuring a maximum around 1700 MiB/s, which I think is due to the drives being connected via the RAID card.

If I wanted to reconfigure the enclosure holding the NVMe drives to be direct connected using PCIe to the the system, is it just a case of swapping the backplane for the x4 version P39780-001 and connecting with cable P53285-001 to ports 5 -> 8 on the system board?

[ngnear]

The current setup with the MR216i-p RAID cards is limiting NVMe performance due to x1 PCIe lanes and RAID overhead.
To unlock full speed, you may replace the NVMe backplane with the x4 version (P39780-001).
Connect it directly to the system board using PCIe cables P53285-001 (ports 5–8).
This will enable full PCIe Gen4 x4 bandwidth per NVMe drive, eliminating the bottleneck.

[support_s]

Hello,

Let us know if you were able to resolve the issue.

If you are satisfied with the answers then kindly click the "Accept As Solution" button for the most helpful response so that it is beneficial to all community members.

Please click on "Thumbs Up/Kudo" icon to give a "Kudo".