Petabyte class SNO

I could install 1PB within a month (then my house starts to get to warm) The cost being about $30,000 …
Could storj.io support this and create a class of PSNOs?

The number of years to fill up that much space is equal to the amount invested. Are you looking for any profits ?

1 Like

Support could be through a “grant” of test data.
A plus 100 of reputation could work too.

Storj said that they want decentralization of storage. The network currently has about 22PB of capacity, so, your node would be about 4.5% of the network capacity. Either your node would get more data than the other nodes (making the data less decentralized) or your node would get the same amount of data as everyone else (more likely).

My node has 4.5TB of data stored on it right now, I do not expect your node to get any more than that. If you get more data just because your node is bigger, my node will suddenly become “much bigger” too.

This was kind-of the case with v2. Due to some weird logic I did not understand, a node would only accept shards that were up to a certain size, that size being determined by the allocated space on the node (so, a 2TB node would accept bigger shards than a 1TB node, even if both nodes were completely empty and the shard size was in megabytes anyway). So, what did I do? Set all my nodes to “8TB”, the maximum, even though they were actually smaller, with a script to change the capacity to the real value once the drive got almost full. Even later I just modded the node to accept any size shard.

2 Likes

The current situation, race for data blocks, is simply exactly equivalent to a race to data centre i.e. a race to centralisation.
My 1PB would be more decentralised as I live in a rural location.
I would expect extra vetting of PSNOs, perhaps even physical, to get the status.

It would still be 1PB in one place, probably with a less reliable internet connection than a datacenter.

At least right now, you need IPs from multiple /24s to get more data than others.

data centre == centralisation

(public | private | home) data center == centralization
Besides that all nodes behind the same /24 subnet of public IPs will be treated as a one node for uploads and downloads. So, more nodes in datacenter - less data across them.
The same will happen with your home datacenter.

We want to have a lot of nodes across the globe, not in a few datacenteres. We should provide high resiliency and durability for the customers’ data.
I hope you understand the point.

OK.
I’ll just stick with 4 nodes. but I might lose interest if there is nothing more to do.

1 Like

Dude, face it as a passive income. Install, configure and let it run…collect at the end of each month :slight_smile:

2 Likes

Not exactly, the official requirements for uptime make it require active monitoring with possible waking up in the middle of the night to fix the node etc.
But this is not mining, there may not be enough clients to fill 20TB let alone a 1PB node.

1 Like

10 data centers in 10 different locations = as centralized or decentralized as 10 home users in 10 different locations

Configuring correctly uptimerobot makes it quite better than mining. Only needing the base monitoring of something happens…at least for my 7 nodes, it’s quite hassle free oftentimes.
But of course, not 1PB!

Both have their differences. While miners seem to require more upkeep in general, in one goes down and stays down for a couple of days, I just lose the income for those two days. On the other hand, while the node requires less upkeep (pretty much only updating it occasionally), if it goes down I have to drop everything I am doing and try to fix it quickly, because 5 hours and I lose much more than just the income for the time my node was down.

2 Likes

10 data centers would be better in this case, since data centers are usually more reliable than home networks.

Yup, absolutely.

I just wanted to point out that there’s no difference in centralization :slight_smile:

Why not 10PB?
if you loose 1PB due to being hardware, or just not having monitoring, or simply being away from home (as some of us are in corona times) - You could loose your entire node and would have to start over again.

The smaller the node, the better it will be, if you don’t want to spend time on hardware or software based redundancy.

1pb comes to about 1.5kW and I don’t want more than that heating the house.
Also it is a comfortable fit for a half height rack.

So no good reason really.

Decentralisation in case of Storj is having different ISPs, Locations, Disk Vendors/Hardware, supply from Power plants etc.
You can have best hardware but what if there be ISP or Power Plant failure in your area then 1PB of data goes offline. Even worse if there be more data shards of a file than everywhere else then files becomes inaccessible until You get online.

That is not how it works.

1 Like