Let's talk about the elephant in the room: The Storj economic model (node operator payout model)

Oh I’m far from a pro in that sense. I only buy expansions if the ROI checks out and with money already earned from Storj. And I use hardware that is already online. One mitigating factor is that I use this HDD to expand an array that is mixed use. Though admittedly the Storj usage has grown a little out of hand. Used space on the array is now 60% Storj, 40% other stuff. But should Storj ever disappear, I will be filling up that space with other uses eventually. So I don’t see it entirely as a sunk cost.

Damn… I want prices that low… mine are approaching 50 cents. Though next year there will be a ceiling price. Not sure what it is yet though. I should probably look that up.

Considering the responses to me suggesting $5 for egress on this topic. I think something like that would cause mass exodus.

Storj is still very small. I would say it’s definitely too early. You want Storj to be a “household” name before increasing prices or lowering payouts. And even then, you have to do it gradually.

So since you called me a pro earlier, let me just say this. I would likely stick around, though I would think twice about buying expensive new HDD’s. It would depend on how the economics work out. But since I’m using always on hardware anyway, my running costs are low, so I could easily stay with the capacity I already have.

Implementing FIAT payouts worldwide is a massive undertaking and may not even be possible everywhere. Plus it would devalue token reserves in an instant. Hurting both Storj and node operators who are still holding on to their tokens. I think the issues you outlined can easily be fixed by better communication around those topics. Plus there are successful web3 customers on the Storj network as well.

And reduce payouts significantly to compensate for that… I don’t think anyone wants that.

I guess the bottom is at $0.30 per TB per month. (Based on 20 EiB of chia)

1 Like

While I understand the sentiment behind this statement, I believe there’s just not enough unused disk space in the world that would be (1) mostly/always online, (2) whose unit size is big enough for the operator to get decent compensation for their time. While it is a good marketing slogan, it’s just a marketing slogan.

Regarding who is or who is not a pro, your classification seems not wide enough. I’d start the pro level at has a rack full of storage, probably at least 5 PB, because at that level with the prices suggested by @Brightsilence profit (revenue minus hardware and operating costs) starts approaching a sysadmin salary.

Pro level would also cover companies like Hetzner, who already operate massive storage at costs below @Brightsilence’s suggestions, and for whom even small profit per-TB would turn into lots of money.

Then there would be some kind of a semi-pro, who has less storage, but still enough to require more than a hobbyist knowledge to maintain and getting revenue big enough that profit is non-trivial (like, maybe you need to do taxes on them?).

Then there would be hobbyists. I classify myself as a hobbyist, maintaining a NAS with 48 TB of disk space, all of that funded from Storj revenue so far. Profit close to zero (i.e. revenue probably covers hardware and operating costs).

Only then there would be casual operators, who would truly operate nodes only on unused storage. But whether it would be worth their time to share that last 10 TB of their small business NAS to Storj, for maybe 20 USD/month, while having to maintain the NAS in likely much better condition than they’d need for their own use? I kinda doubt. 20 USD is around 1-2h of minimal wage in many countries. Besides, that would still require node software to be a lot more reliable and have a lot more features to reduce the effort from SNOs needed—and this is costly to develop.

Of course right now Storj doesn’t have enough data for actual semi-pros, let alone pros. But, frankly, I wish I could retire with a nice rack of hardware giving me some bonus pension…

2 Likes

It’s pretty much the same as what Storj is doing now - filling the nodes up with test data. Node operator gets paid and Storj can delete that data to free up space if needed.

And then there would not be enough resources for the network.

I do not know how many people have:

  1. multiple large drives that are in operational condition, but unused (either completely unused or lots of free space)
  2. skills needed to set up a node that works properly
  3. the desire to do so and to mess around with exchanging tokens.

Let’s say that someone only has small unused hard drives, say a few TB. Would they be setting up a node just to get $1/month? Maybe, maybe not, but there would definitely be fewer node operators and much less space on the network.

Welcome to Storj v2 requirements :slight_smile:, at least officially as I remember. In practice they were more harsh than the current ones.

While I think that 5hours/month of downtime is way too harsh, your proposal is in the other extreme.
With such low uptime requirement, there would be huge problems for customers to access their data. Sure, if the nodes went down randomly, maybe it could work, but if a lot of nodes go down at the same time every day (end of work, turn the PC off), then those nodes would be pretty much useless.

The way I see it end up is that Storj only works for scale operators or small operators with low to no operating costs. Right now it might still be profitable to build a system and earn back the investment, but long term, that might be tricky. ROI may simply take too long for dedicated hardware at small operators. But that’s why it’s so important to know what the future will look like for payouts. I can make my suggested numbers work, if I know long enough beforehand that that is where it’s going to end up. But I would have to think a lot harder about hardware investments.

1 Like

That model still works though. If you start with 10TB free space, like I did. By the time that’s full you’d have made $300. Plenty to buy a decent expansion (16TB around that price). And it only becomes easier to expand. Because by that time your original node is making $33 a month. So it wouldn’t take too long to be able to buy a new HDD again especially if you add the income of your second node. You could afford expansions faster than the HDD’s fill up. And all the rest is profit.

You then have 26TB of space, which takes 4.5 years to fill up and you make about $2500 in that time, while only ever spending $300 on a single HDD.

2 Likes

It takes 3 years to fill it up. By the end you’ll be making $63 per month. meaning you could buy one of those every 6 months roughly. But yeah, that’s assuming payouts don’t drop significantly.

More importantly, it takes about 21 months to earn back the initial investment. But profit is quite good after that. In month 22 you already make about $40 a month.

1 Like

A little out of topic but my thumbs are itching (I’m on iPhone).

The STORJ token and the web 3 marketing are not just beneficial. It is arguably the competitive advantage and the only reason STORJ comes this far. Here is why.

Let’s start with some facts. First, market capitalization, in USD.

STORJ : 195 million
Backblaze : 133 million

Now, look at stored customer data.

STORJ : 13.7 petabytes
Backblaze : 2,000 petabytes

Backblaze has more than 100x of stored data but much less market cap. This, my friends, is gold.

The reason STORJ can subsidize node operators and at the same time undercut competitors’ prices is because of the amount of reserves which is enlarged by the hefty token valuation.

And how come the valuation is so high? Because it is all about crypto, web 3, decentralization, and basically… the future. And because there are “huge amount of idiots” who believe so. AMOF node operators should thank them. Your subsidized payouts are at least partly from these people.

Without these things, STORJ is just a super small and super depressed version of Backblaze.

You can replace Backblaze with any comparable web 2 cloud storage provider. Same conclusion.

3 Likes

Uh, not exactly. Storj does not guarantee test data existence in future. The contract would guarantee some allocation to profit regardless of usage, and hence the risk for a SNO who manages to snatch this kind of contract is smaller. Auction would mean though that there would be a limited number of contracts at prices lower than regular revenue, making it a risk/reward trade-off compared to just sharing disk space to customers who can’t plan ahead much.

Heh, no, I started only when Storj was close to production use, and missed most of the early bonus payments. @Brightsilence, and many other folks here were earlier.

Oh, no, this is definitely not the only selling point for Storj. For me the biggest thing is acting as both storage service and CDN for a relatively low price. This is absolutely huge, much more important than just utilising unused storage.

1 Like

Yeah. Point taken. I mixed those up. Me making the reference was unnecessary actually.

1 Like

I’ve spent $3000 on my storagenodes for ~$40 per month. I don’t think that makes me a “pro” as BS said above.

5PB in a rack? meh! You would maybe if you world’s unluckiest sysadmin replace a disk each day. Is that a full-time job? I think that maybe the case is for commercial partners but if not…I’m in where do I sign up?

P.S. Mr BS, you may not be a pro storagenode operator but your attitude is the most professional I have seen

4 Likes

(speaking purely as a storage node operator, which I was before I got paid to work on the code)

Keep in mind that I already had a server and beefy router and I just enjoy maintaining them and hosting things.

My marginal costs for disks only are about 0,40 €/month/TB. 75% of this is electricity and 25% wear/depreciation. Unless the price falls below this I have no financial reason to leave the network. But at such rate it’s not reasonable to add another disk to my existing server.

My server and router without disks both use about 60W which comes to €45 of overhead per month (the hardware is pretty much free). And without the bandwidth need of the server I could save €20 at my ISP.

Amazingly the Storj payouts cover most of these costs. Also some of the electricity heats my home.

We don’t need to guess how low prices can go before shortages arise. We can check with the competition on the free market that it is around 0,73 $/TB/month. Probably for Storj a market price would be lower because Storj requires no collateral and it requires less CPU.

So to be clear I am not saying that (1) any node operator specifically will not leave the network at $1 or (2) Storj should lower prices to $0,73. Just sharing my perspective that I think the network could be run at such a price and I personally won’t exit at such a price.

4 Likes

Leave at $1? Maybe not. Start a new node (either to add capacity or because a node was DQd) or expand the current one? Also maybe not.
Unless the demand increased enough to compensate for that.

Everyone has the minimum payment threshold below which it is no longer worth the effort. Payments drop below that value - people quit or just stop looking after the node and it eventually crashes.

I have a big server, so my fixed costs are high, but marginal cost, especially egress, is low. If the rates go down, but traffic goes up I’d be OK with it. If rates go down and traffic stays the same, then my only hope is that the token price goes up in the future.

1 Like

Please note that some of the most consistent contributors that bring egress to nodes on the network are in the web3 space - customers such as POKT and ANKR - the former you can read a case study about here - these companies provide blockchain synching at speeds they could not achieve without Storj.

Other customers in the web3 space are NFT projects such as Ultimate Division and Boonji - you can find case studies for these 2 also on the same page.

I am not quite sure what you are referring to with ¨Token Marketing¨ - Storj Labs does NOT do any marketing related to our STORJ token. In fact, we constantly correct people on various social media channels who want to look at STORJ as anything other than a utility token used for paying our node operators for services rendered, and available to customers to pay for storage and bandwidth.

8 Likes

I just meant that I fall into the idiot category. No offense

1 Like

Lets not dwell on this. It wasn’t meant to offend anyone here, but addressed to people who think Storj data is stored on the blockchain. Even then I wouldn’t call anyone an idiot though. I’ve seen Storj partners describe it as such as well as a lot of media. People aren’t idiots because they simply got presented the wrong information. That said, you were definitely not the target of that remark.

I appreciate that. Thanks. I try to be fair in these discussions and while this topic may have gotten ahead of the actual planned changes, I think it’s important to discuss this. Because at some point this ball is going to drop, even if it isn’t immediately with the upcoming changes to the economic model. Since we’re talking about multi year ROI. If it happens within two years, it would still impact investments node operators make today.

I see the CDN aspect as something underrepresented in this discussion. Are there comparisons to CDN services? It seems like a decentralized CDN would have the biggest advantage in that market. What place does CDN service have in Storj’s future?

Hello everyone. I’ve been a SNO for 2 1/2 weeks. I had a thought that could reduce costs for the network: node downtime for maintenance. If this has already been brought up in the past then please ignore and it just proves that I’m not original!

The basic idea is to allow for shutting down a node for maintenance, maybe specifying an estimated amount of time, and the network won’t immediately consider all of the stored data as offline and that the data might need to be moved to another node where all that egress has to be paid for.

I started my node and then a week later decided to move my server to another room. While it was down I rearranged some cables and tidied up, and then plugged it in and turned it back on. It was down for probably about 2 hours or so because I had to run to Microcenter and get another ethernet cable, but should have been closer to 30 minutes. (I should have planned better!)

I don’t remember the exact amount but while it was down the trash went up by about 0.5GB, and of course I am only storing a small amount of data right now. Somebody was paid for that egress. However, if I could have told the network that I’m doing temporary maintenance, it wouldn’t necessarily have had to pay for egress of that data. Maybe after 3 hours if my node was still not up, then data can be moved to other nodes. You could play with the numbers.

Obviously there are things to consider, what if multiple nodes go offline for maintenance in the same time window, what if the planned window was 30 minutes and it was really 2 hours like in my example, what if the plan is for 15 minutes and the user drops the hard drive down a flight of stairs, etc. But for hopefully simple things like I’m updating my operating system, or maybe shutting down to attach a new hard drive, it doesn’t make sense to incur costs for that.

This would incentivise node operators to shutdown with the flag because they could retain the data, their income, and Storj is incentivised because they would not have to pay to move data around.

1 Like

The idea is good and was proposed in the past too. But it is also dangerous, better to let the system permanently repair pieces.

Also a maintenance should not take more than 5 min, of course, shit can happen while upgrading / reconfiguring etc… in such cases is ok to lose some storage.

I don’t think the increased amount of trash was not due to your node being down. 0.5G of trash isn’t really that much, it will come and go in larger chunks than that just from the normal operations of the network’s users.

A little bit of downtime like that will hardly be noticeable by the system. Enough parts of the customers data will still be available so they won’t notice either.

A repair is triggered only when the number of available pieces become too low, I think around half of the nodes need to be offline more or less permanently for that to happen.

1 Like