Let's address the /24 subnet limitation ipothetic removal

[shka]

@Knowledge

I tend to think if we got rid of the /24 filter, that the majority of nodes would get less data, not more. Even if you are splitting data with multiple nodes, the fact is that data centers have large pools of static IP’s that they could assign to each node. They would blanket the network with far more IP’s than the home users would have. That’s the entire reason for having the /24 rule in the first place, to prevent this. They would end up with the majority of the data, and thus it would centralize.

Is it better to support IP types except for the data centers? For example, www.ip2location.com tells your IP type as “Usage Type”. Then even if /24 filter is removed, it’s impossible to use VPS/VPN through the (honestly registered) data centers. But of course, it allows residential-proxies/-VPN still.

[Pentium100]

ip2location does not always have accurate information, the “usage type” is not part of RIPE data or anything. One ISP had its IPs designated as datacenter until customer complained about it and then they contacted ip2location to tell them to fix the assignment.

[EasyRhino]

As mentioned before, I don’t see a difference between a /24 and a /32 from a reliability perspective. a /24 isn’t really that big, only 256 addresses. It’s probably still in one data center, or in one neighborhood, and vulnerable to network outages.

I like an idea kind of mentioned earlier, that works within existing storj concepts. and that would be:

for additional nodes in a /24 (or maybe even something bigger like a /22? limit the bandwidth for nodes in the vetting period but after they are vetted, release the restriction.

[Alexey]

this limit is already happened automatically: old and new nodes selected one from /24, but also a new one has a separate pool of 5% ingress from the customers, all new nodes shares it.
So, 2 new nodes will get half of the data share each from 5% selection for their /24 subnet.
Or maybe I did not get your suggestion?

[humbfig]

No, no no…

First, decentralisation is important. Otherwise Storj is just another storage provider.

Second, people who think they will get more traffic if /24 limitation is removed are wrong. Everybody is smart and will deploy a bunch of nodes. Collectively, they will all look dumb.

Third, no deposits upfront. Just NO!
If you had to make a deposit for your first node, the Storj model, with amateur SNO’s with a bit of space to spare, would never work. I would not have started a node if I had to put money into it. Now that I have a feeling for the thing, I could do the math and decide, but for a first time SNO it would just not work.

I think only the “semi-professionals” come to the forum to discuss these things. I hope Storj management realizes this and doesn’t forget what Storj is about. A bunch of “amateurs” with a bit of space to spare ensuring true decentralisation (even though the satellites are a stain on the decentralisation, which should be better addressed).
Storj is not for semi-professionals. Their specific interests should be disregarded.

PS- Notice that I’m speaking against myself. I regard myself as a semi-professional. I run 6 nodes on 3 different IP/24. Truly decentralised, since I have only 1 IP at my home (and no VPS, VPN…).

[Krimarai]

Held amount is a good system. I don’t think that deposit is better. Held can just be higher if there is a need to remove subnet limit.

[FREDY]

I agree this /24 limitation should be reviewed and may be an overkill to think data would be loss if something happens with those nodes behind that specific /24.

To start with there are broadband service providers that will have at least 255 different clients on different physical addresses behind a /24 which is already fair enough for diversity of environments, even considering a neighborhood, city or even region which is mostly the case.

Also one thing that needs to be reviewed is the need to run a different storagenode for each new disk you add which makes this scenario more restrict unnecessarily. If one has multiples disks at home let them share them with most efficiently possible and avoid making it less attractive to store data. Having more and more people willing to share resources is what should matter for a even robust Storj diversity, capacity and lower costs.

[Alexey]

@FREDY

If one has multiples disks at home let them share them with most efficiently possible and avoid making it less attractive to store data.

The storagenode software should not replace the system software. If you willing to use RAID you may do so, even if it’s not needed for the network, see RAID vs No RAID choice.
With /24 filter and ability to run several nodes they all work as a big one node, exactly what you want. In case of disk failure you will lose only this small part of your common data, not all as in case of RAID0.

[FREDY]

Hi @Alexey the point about using RAID in a device with multiple disks is quiet common specially in a local device with multiple disks that ease up things including disk replacement and I see no issues in that large RAID (on the top of a RAID5, 6 or even 10) be seen as a single disk to Storj network as long it is well maintained. I do understand that for a network system may not be needed, but in the other hand is not something that can cause a harm.
In fact the use of RAID (5, 6 or 10) in a local system avoids any single disk and amount of data be lost and a more complex process have to be conducted if there is a much easier and flexible solution available. If disks are used individually for each storagenode process some of them will be lost at some point and it is in analogy similar to loose a portion of a RAID 0.
RAID 0 should never be used for multiple disks in a array to serve Storj.

Correct me if I am wrong, but when the use of RAID is generally dissimulated and the provision of multiple storagenodes and there is this /24 restriction I understand the satellites may avoid adding more data into that ‘group’ of storagenode where that could receive more data and act as a more capable system.

[Alexey]

The satellite just trying to make sure that each /24 subnet did not get pieces of the same segment, it doesn’t limit how much data it could receive.

@FREDY

If disks are used individually for each storagenode process some of them will be lost at some point and it is in analogy similar to loose a portion of a RAID 0

unfortunately not the same. In case of one disk failure the RAID0 is totally dead, but separate nodes will survive. But better to do not discuss RAID vs No RAID again here, if you want to - we may continue in the related thread.

[andrew2.hart]

Subnet	Nodes in subnet

1 65.21.93.0 1380
2 74.208.76.0 244

How do this work? Is that one user?

[Knowledge]

I believe an IP that ends in a zero is usually some kind of gateway device. So, maybe a pool of users behind a proxy, VPN, NAT Pool, VPS, or similar.

[donald.m.motsinger]

First IP belongs to Hetzner, the second one to IONOS.

[snorkel]

The exact IP is not ending in 0. These 2 IPs are displayed on Storj Net Info, as subnets. The 0 is just to show that it is the entire subnet.
From the number of nodes behind it, it makes sense to believe that the entire subnet belongs to one entity. And… Hetzner it is

[snorkel]

I wondered that myself but now it hit me: they use the cloud server for full nodes, because the ingress stopped, and you don’t care anymore for ingress shared with others when you have a full node. So they pay maybe the 4.52€ per month for the cloud server, and keep the full nodes online, freeing the much needed personal IPs for new nodes. In a few years, we will be all Hetzner clients .

[donald.m.motsinger]

I think Hetzner and IONOS might do it by themselves to earn a “little” money on the side and use their unused space. If it were customers of them I think it would be spread more around their IP ranges. In Hetzner’s case it would be 65.21.0.0/16.