Hi, I have 5 fiber connections with a local provider in Italy that provides multiple public IPs on different subnets on the same fiber connection. In fact, 1 week ago I requested 4 more IPs which will be assigned to me shortly. The cost of this service is truly trivial, €3 per month for each IP which pays off abundantly with medium/large sized nodes. Obviously the 5 fiber connections located not only serve the nodes but also provide other services for my work. Today I am already at 120TB shared with around 100Tb already filled. By next month I will move to around 200TB with the migration to new HDs that are arriving.
1 Like
Hi, but this control interface what is?
I believe - Grafana + Prometheus with storagenode-exporter
May you share a link to the dashboard software?
What HDDs are you having and how they perform? What are the new ones? Do you have UPS+write cache enabled?
I Buy only WD Enterprise DC HC SAS. These disks performance is about 230MB/s . I have ESXI over Hp Proliant DL380 G8 with 3/4 nodes for one VM over one IP. The VM is on SSD and the DBs are on SSD. I have 48GB Ram withe 32GB Ram dedicated to PrimoCache for L1 Cache
1 Like
Here another Italian SNO.
I own two nodes under one IP, FTTH connection.
I get about 1.5TB per month!
I own a truenas server with ryzen 5600x and 64 gb ddr4 ecc ram.
The first node is running on truenas container, the second in a ubuntu server vm.
Since I am replacing my 8 x 4 TB WD RED Plus with Exos x16, I am reusing the WDs for storj nodes. Connected via usb. (The second node in vm is the first hdd replaced).
It is exactly my already online hardware not in use 
The hdds connected via usb are placed in a stackable 3d printed caddy.
You may run as much nodes, as you have HDDs, the VM is a waste of resources in my opinion.
1 Like
I prefer to avoid mounting external hdds in ext4 on truenas. It’s an appliance OS.
I avoid a thousand problems, including failure to boot in case one breaks.
The resource impact on such hardware is about 1%, negligible.
There’s a mount option called nofail, taking care of that problem.
Besides, as strong as stated here that you should include the mounts in your fstab is in my opinion a bit too strong.
I actually use LXC-containers, so the so-called subsystems have the disks in their configuration and fail to boot as soon as the disk isn’t there. No big deal, because designed that way.
As a Dutch SNO, I have 5 IP-addresses. I likewise get about 1-1.5GB nett ingress per IP, as long there’s enough space left.
I’m quite appalled by the fact the TS apparently took 27 months to fill up 10TB. I’m around here since April. I increased my storage over time to 30TB. It took me less than half a year to get 10TB and les then 3 months from 10>20TB.
1 Like