Poor Norco... I have 2 RPC4224 cases and the backplane is slowly decaying. I have 2 slots defunc now. Luckily I didn't fully populate all 48 slots so I can shift around a bit. But who knows what will happen if more and more slots go silent...
Back then I got Norco because they are cheaper than...
Thanks for the reference! Seems it's a hit or miss thing. When the time comes (probably next year) I'll test more to see it myself. Also it's gonna be ksmbd so even more unknowns@_@
NVME-oF is something I excluded because it does no help with latency, either via RoCE or IB. I'd prefer my storage being local ZFS with local NVMe drives (can be passed through).
I checked Hyper-V and it seems its virtualization offering is rather limited, cannot cover a lot of edge cases, e.g...
Proxmox and napp-it cs are new to me. Need to research more on it.
Any reference to Besterino's effort?
If using Windows, would you recommend using it as a host with Hyper-V (I only indirectly used it before with WSL2) or as a guest in ESXi/Proxmox?
I have been following ksmbd for a while now since it seems to be a silver bullet to solve my problem.
My proposed use case is all-flash ZFS all in one with ZFS exposed to ESXi (and VMs) using NFS and Windows client machines using SMB, utilizing 100Gb RDMA IB (and 100GbE for regular IP traffic)...
Ah thanks! That's exactly what I wanted to know, and PLP is something I missed to consider. The cost is ~$150 on eBay so acceptable.
Just wondering has anyone used it for ZFS SLOG? It seems too good to be true...ending my decade long effort of finding the perfect (and cheap) ZFS SLOG device @_@
Hi all,
I recently come across these devices called Radian RMS-200 and Radian RMS-300. Looking at the description and spec, it seems to be a "spiritual successor" to Zeus RAM and a perfect fit for a ZFS SLOG device. However Radian does not specify the latency of it.
On the other hand I already...