NVMe/fc on aff with just ssds and vice-versa

[elic_co]

Hello everyone,

when using nvme/fc on an A300/A700/A700s for example with just ssds as the backend, what is the performance beneift/degregation?

Is ontap wafl converting nvme commands to scsi just before writing them to the disks?

and if my application is mostly write intensive does this mean I will get "hurt" less becuase most of the nvme commands will be written to the nvram anyway?

And also from the other way around, if I use nvme disks in an A320/A800 but with fcp for example as the protocol, does iops get translated from nvme to scsi and vice-versa depending if they are read/write ops?

Thanks,

[Alberti]

With NVME you have around 30% more speed above the protocol.

[andris]

ONTAP’s initial NVMe-oF protocol was NVMe/FC which was very easy to migrate to if our customer was already using FC, which the majority of our block customers are. We have several NetApp Verified Architectures (NVA) with reference architectures and documented test, configuration and performance results that can be easily replicated by reviewing any of the following:

• NVA-1126 - NetApp Verified Architecture: NetApp and Broadcom Modern SAN Cloud-Connected Flash Solution - Oracle and SUSE NetApp Verified Architecture ...
• NVA-1127 - NetApp Verified Architecture: NetApp and Broadcom Modern SAN Cloud-Connected Flash Solution - MongoDB and SUSE NetApp Verified Architecture...

• NVA-1136 - NetApp Verified Architecture: NetApp and Broadcom Modern SAN Cloud-Connected Flash Solution - VMware vSphere Verified Architecture Design E...