Title: [Inquiry] Revenue and Network Performance for Deploying Storj Nodes in Yunnan, China (Southeast Asia Region)

Node Operators
1

Category: Node Operators

Content:

Hello everyone,

I am planning to deploy a set of Storj nodes in a data center located in Kunming, Yunnan, China, which is in the Southeast Asia region. The infrastructure is stable, and I would like to seek advice from the community, especially those who have experience running Storj nodes in China and Southeast Asia.

My Infrastructure Details:

  • Location: Kunming, Yunnan, China (Southeast Asia region)
  • Storage Capacity: 12PB (planning incremental deployment with 20–100TB per node)
  • Public IPs: Up to 10 public IPs available (each node has an independent IP)
  • Electricity Cost: ÂĄ0.38 RMB/kWh (approximately $0.05 USD/kWh)
  • Bandwidth: 300 Mbps down / 30 Mbps up, very low bandwidth cost (ÂĄ360 RMB/year, about $50 USD/year)

My Main Questions:

  1. What is the realistic revenue range for deploying Storj nodes in China and Southeast Asia? How does the upload bandwidth limitation affect revenue?
  2. Are there any issues running Storj nodes with a Chinese IP, such as failed audits, high latency, or uneven data distribution? How do these factors impact earnings?
  3. Considering the 30 Mbps upload bandwidth, is it better to run multiple nodes simultaneously or expand a single node for better revenue?
  4. Are there any success stories or case studies of running Storj nodes in China or Southeast Asia? Any tips or recommended tools for node management and monitoring?

I am a developer in the smart home and IoT field, familiar with server management and network operations. I would appreciate any insights or suggestions that could help me optimize my node deployment. Once everything is up and running, I will be happy to share my results and experiences with the community.

Thank you for your help and sharing your experiences!

1 Like
2

You’ll probably never fill up 12PB. Storj currently has only 28 PB total customer data (Grafana) and this number includes the Select network. i don’t see how you will be able to hold half of the total network…
And I think I have just recently read that some other SNOs fill rate is 2 TB per year per node. If you have 10 IPs this could mean 20TB per year if the IPs are in different /24 subnets otherwise all nodes in one subnet count as one node altogether.

2 Likes
3

The maximum data you’ll realistically fill is 8TB/node (if you only run a single node per /24 subnet). That’s the projection for the next 5 years. The current download is ~10Mbps (on a good day) per IP-in-different-24-subnet (that’s without anyone else running any node on that /24).

That’s 3TB/month/unique-node-per-24-subnet.

Want my honest opinion? Don’t even bother spinning up a single node.

1 Like
4

Here it is: I create videos about My nodes - #67 by HGPlays

3 Likes
5

Yeah… 2-3TB/subnet/year… across (presumably) 10 subnets… would have a high estimate of 30TB/year? So if that could be sustained… 12PB would only take 400 years to fill?

Or am I doing the math wrong?

Spend less on storage space, and more on IPs. Like get to 200 subnets and fill 2PB first.

6

With 300 Mbps download speed one can never fill 12 PB. Deletions of existing data would offset incoming new data much earlier. 100 TB might be the maximum. After some years…

2 Likes
7

On the topic of sharing: Th3Van shares his progress and build. You’ll see his notes go back to 2020… and up until now he has managed to store around 650TB. So that’s 4 years to fill around 5% of what you’d like to bring online.

8

There is also the location to consider. I would think that a storj node in Asia gets less traffic than if it were located in Europe or the US.

1 Like
9

Low speed, 10 ip.
I think you can get more stable income by selling this disks as new, like an other China fellows :grinning_face:

4 Likes
10

Let me further clarify the deployment details:

  • Our data center is located in Kunming, Yunnan, China, considered part of the Southeast Asia region. We have stable operations and maintenance capabilities.
  • We currently operate 120 servers, each equipped with 96TB of storage, with a total capacity of over 11.5PB, scalable to 12PB as needed.
  • Each server can be configured with independent bandwidth, and there is no limitation on bandwidth. We can flexibly adjust bandwidth allocations based on Storj network demands.
  • Each bandwidth line can map up to 10 public IPs, and a single server can be assigned multiple bandwidth lines to meet subnet separation requirements for Storj node deployment.

We are looking to better understand the Storj Commercial Node Operator Program, including revenue potential, deployment strategies, and official support for nodes hosted in China. We welcome advice and guidance from the Storj team and community members.

11

You should fill this form.

2 Likes
12

This clarifies your initial post.
I don’t think there are CNOs here, at least I have not seen a user outing himself as such.
As a SNO you operate and get paid differently than the CNOs. Maybe this helps you with some additional information:

13

The big guys serving Storj Select have dedicated support and dedicated channels; they don’t spend time on this forum anymore.
Your setup is more suited for Storj Select; you should contact Storj Inc directly.
Another thing to consider… activity on the Asia server is very low.

14

I Can confirm this is roughly what I have seen on each subnet I have access to, where I am alone.
About 2TB yearly.
But it’s important to keep in mind a lot of deletes has happened, and also customers needs can change at any moment.

I will keep expanding if:
Drives get full = upgrade to bigger.
My internet deals expand =I have access to more subnets.
Other than that I’m not deploying anymore (unless I find a cool video idea for YT) :grinning_face:

1 Like
15

You should forget this idea in China, not profitable in China. That is why you only see few SNO spots on the statistic map.

1.I agree there is low electricity cost and low rental fee in China, especially the Yunnan part.

2.However, the income you could receive is also very low:
a. Four core official satellites are sitting on Google IPs. You have to do partly routing masque. You cannot rely all the traffic throughput on VPN, otherwise the bill must be higher than your revenue.

b. Most customers are in US. Traffic flow is very slow across the sea. So, your harddisk are just idle and burning electricity. Unless you pay special direct international lines, worth to do so?? The key point is the line latency especially in stateful TCP protocol, not QUIC UDP. The bandwidth upper limit is not too important.

c. Very few IPv6 customers data, most are IPv4. IPv4 lines are also expensive in China.

I am in HK and test this for few years. The revenue is just break even. The most crucial factor is not electricity. It is the ISP line fee and type of customers. For your reference, about 200KBytes/s flow in average, even it is a 10Gbit/s line.

However, if there are lots of Singaporean / Japanese customers, my disks could be easily fully filled in short time. Then, i could immediately reuse the same IPv4 and add another disk. The utilization of that ISP line is high in such scenario.

I won’t imagine there are lots of Chinese customers because of the Google routing issue. Also, those Alibaba,115, Tencent …storage services are already very cheap. Storj is not competitive in this country.

You better do a pilot test first, with an accountant.

2 Likes
16

This is an illusion. It will always take years to fill a decent disk from public network. Select network might be different.

17

We do not have AP Select, only AU Select, which in my opinion is far from China, so may be not so good as the public nodes in China.
But Google routing issue could be a blocker, I have to agree.

I hope that we can use our own compute to host at least part of the satellites, and due to a distributed nature of our compute offering it may help to overcome this issue too.
The only limiting factor I think, that we need to use only SOC2 Type II compliant compute nodes.

1 Like