Recent content by kapone

I just gave you a real life example of handing 2.5gbps symmetric fiber internet with el-cheapo hardware (and software). Does it need it to be cheaper than that?

Nope. You can do that too. Depends on what you're trying to achieve.

Any system from the last..oh..say 15 years or so (essentially Ivy Bridge onwards...in Intel land) will be less than 50w at idle with those specs. Hell, an X10SRH type board with onboard SAS and a 8 core CPU idles at ~40w with a platinum PSU.

p.s. Talking about software...the combo of Starwind CVM (Linux VM based on Ubuntu 20.04 LTS, I believe) on Proxmox (I'm using the latest 8.1.4) is sweet. Does ZFS, ISCSI, iSer, NFS, SMB, NFSoRDMA (still gotta test this) and NVME-oF (soon to come) out of the box. You can literally create shares...

@kevindd992002 - With all due respect...it doesn't look like you have enough expertise for these type of things. Asking on forums is fine, but all you'll get is personal experiences with said piece of technology or hardware. It takes years and years of real life experience to be able to...

Yes. But...too many variables. p.s. I do 2.5gbps FIOS with many static IPs...on a $50 system. This is my "core" high density box. Three systems in a 1U chassis. - OpnSense - Router/firewall - DHCP/DNS/Adblocker server - A few critical VMs on a small Proxmox server.

Well aware of them. If I had to use those...I might as well use the Delta 1212SHE 38mm fans (which I do use). Even at 50% PWM (~1400rpm) they're close to silent and push way more air than the Noctuas.

You could...but you shouldn't. The 847 can hold 36 LFF drives...the Noctuas don't hold a candle to cooling that many drives.

The 847 can be…tweaked…to be quieter. I use the Norco fan wall to put 120mm fans it, but there are other ways. With SQ power supplies in it, it is no louder than an 846 (that I mod similarly).

This is exactly what I mean by "arcane commands". So...some repo on GitHub says...do this. Really? No, I'm not gonna do it. Either I'm gonna figure it out the hard way, or wait till the PVE devs get their **it together and actually implement it right. All of this is gobblygook, it is...

On a side note, don’t even get me started on PVE’s management IP setup… It will grab an IP from the DHCP server during install, just fine…and then use it as a static IP…and not really tell you… What brilliant mind thought of that? And if you really wanna use a dynamic IP for...

I‘ve been watching this with mild interest for a while…but honestly? I don’t get it… Are we playing with Legos now ??

For data centers and hyper scalers and the likes of AWS, Azure, GCP etc, yes. For the average SMB or homelabber, 40/56gb networking is still...well..rare. Problem is the infra, not just the nics and the ability to configure and manage it on an on-going basis. You see a fair number of 10g...

Update 2: The OVS issue turned out to be a blonde moment on my part and not getting the config right. SR-IOV works just fine with an OVS bond/bridge with the CX3 adapters. Oh, and RDMA on the VF adapters (Linux guests only) works great as well with PVE.

That's perfectly fine. If you don't need/want it, you don't. But...:) In the era of storage devices becoming extremely fast... *cough* NVME...a fast network fabric is almost a necessity if you're doing anything more than typical home networking. I realize SR-IOV/RDMA/ROCE is not for everybody...