How many node do you have?

I can actually imagine a large data center using their spare, unused capacity for services like Storj.

Consider, let say, Hetzner. They probably have tens of thousands of customers, so likely manage a similar number of IP addresses. They can probably chip off one address out of every /24 block they have access to, because most customers only manage single IPs or maybe some /28, /27 subblocks. They have plenty of storage that is waiting for customers… while unused, why not use for Storj? And they can do it on scale, meaning a single person automating hundreds of nodes.

Now the work involved wouldn’t probably pay off the engineer tasked with setting up the nodes, Tardigrade’s too small for that yet. But I wouldn’t be surprised that if Tardigrade manages to take off, most of storage will be provided by places like that.

Most of it can be automated easily for their scale. Kind of like renting amazon spot instances back in the day, hetzner would use their idle storage for this easily.

Is it worthwhile? That’s another question, i’m not sure the demand is quite there yet.

1 Like

This makes me glad there is no datacenters remotely near me. Cause you know people with VPNs also hide there true locations…

Doubtful that any datacenter would set up nodes to earn extra profit. Compared to typical web applications that’s a heavy load on HDDs which might already make the extra income irrelevant because HDDs die earlier.
Besides, datacenter owners aren’t interested in crypto stuff. They have better things to do than following all developments in the crypto world. Maybe there is some owner who finds storj but then it might be a single datacenter in a whole country.


Hetzner’s already offers large storage servers at approximately 1.6 EUR/non-redundant TB (before VAT), so less than what can be earned from Storj. Given that this is what Hetzner’s asks, it must already cover running and maintenance expenses and generate some profit. So it would be as good as having actual customer paying for this server, or maybe even better. If only there was more data stored…

No, but Hetzner’s in storage business. Storj would essentially bringing them customers. They might not care whether customers access their storage via FTP, NFS, SMB or Tardigrade… why would they?

I dont think hetzner doing that as its make lot of stress to storage, maybe it would make profit but only in long term and its not profitable to run enterprise storage for storj and for that count of nodes need additional hands to update nodes time by time so not sure.

The enter a new point of error via Tardigrade, so they might actually trust their own hosting solution more than Tardigrade

Anything is better than SMB, and Hetzner’s offers it.

1 Like

5 posts were split to a new topic: Run NFS instead of CIFS it runs like a charm

But to stay on topic :slight_smile:
1 RPi4 node with 3.6TB in RAID5 iSCSI that will expand in the time my data grows current setup can grow to 48TB

Regarding RAID5 and today’s disks:

You risking to lose those TBs at once

I know the risc of RAID5 but is better then nothing. Raid 10 is way better but I loose volume. If a drive fails I get a notification and I can replace it directly. Although maybe my array is damaged then I need to start over. But there are allot of single nodes or even raid0 nodes the have a higher risk of failure.

Next to that if you use enterprise grade drives the change of failing on a RAID5 is zero to nothing.

Cool calculator for this:

You also can read this thread:

The SOHO drives have a URE = 1e-14, Enterprise grade have URE = 1e-15

Migrating to raid 10 as we speak haha!

I prefer raid6 over raid10. Raid6 can handle the failure of any two drives at once, while raid10 may handle the failure of up to half of the array at once, but it can also die with only two failed drives.

Raid10 has better performance though.

I beleive that ZFS’s raid5 equivelent (raidz) solves some of these problems.

those are some really bad failure rebuild rates for raid5