We had these discussions years ago, concluding that if Storj succeeds, they will need several orders of magnitude more nodes. No technological solution survives this kind of scaling unscathed, and we have already seen growth blockers like bloom filter generation process requiring nontrivial rework.
Out of curiosity—I’m not visiting such places. How do these advertisements look like?
Node operators are a commodity. Per economy’s natural laws, price of commodities closely follow costs with barely any margin. The fact that Storj Inc. maintained their current payment structure for so long is admirable, but justified only by the need to maintain a reasonably stable reserve for growth, that is, effectively conditions that can change any time.
Besides, I believe Storj can absolutely be swimming in nodes. It’d be enough for them to implement a tit-for-tat offer, we have discussed some of these in the past too. My understanding is that the main reason they are not doing so is that it doesn’t seem necessary yet, so no point in spending engineering time on it.
Yeah: we saw back in the load-testing days that people are willing to bring as much capacity online as the market could ever need. I bet a lot of SNOs think “…if Storj fills a drive: I’ll bring another online…”.
I don’t have a fancy setup: but it will never fill: space is easy to add.
My costs have decreased the last few years. Peak electricity costs was about 0.40€/kwh in 2022. Thanks to hashstore I was able to cut my power consumption in halve and keep operating my nodes. Since then my electricity costs have lowered back to normal 0.30€/kwh.
In addition I have also installed small solar panels on my balkony. It will reduce my electricity bill by about 100€ per year. So a very small solar installation but it will still pay for itself after about 5 years. Solar panels have dropped in pricing over the years as well. A replacement would now take just 3 years to pay off.
Now I am looking into AC battery packs for “arbitrage trading”. Charge the battery when there is too much solar and wind energy in the grid. Consume it when the price is high. With a 5kwh battery I could lower my average electricity costs down to 0.20€/kwh (conversion loss already included). Right now these battery packs are too expensive to make that work. It would take about 15 years to pay off the battery. I need cheaper batteries and better politics. (It would work with DIY battery packs. They are a lot cheaper but they take up more space that I don’t have in my apartment)
Long story short. Electricity costs have not increased for me at all. They are slowly decreasing over time.
Same for my bandwidth. After years of waiting I was able to upgrade to a fiber connection. Like 10 times more bandwidth for the same costs with the option to go down 1 or 2 steps. I am saving about 120€ per year compared to my old internet connection.
And last but not least hardware. Last time I had to buy a replacement HDD it was about 15€ per TB for a refurbished 20TB HDD. That seems to be stable over the years.
So no I don’t see an increase in costs. I am happy with the current payout rate.
Guess what. I raised that question internally. Even if storj would lower the payout rate by 10% I would still be happy running my node. I could tollerate up to 20% lower payout rate. So 1.2-1.35 per TB used space. As a node operator it hurts to ask but as an employee I had to share my numbers internally.
It was considered but for the time beeing dismissed. The potential cost saving is relative low compared to other ongoing costsavings. First time I like the low priority answer
A storagenode works great with refurbied HDDs. There is no need to buy factory new drives. You are still correct. Even refurbised drives are at 20-25€ per TB right now. I have seen 15€ per TB earlier this year. I am optimistic that we will get back to that price level later this year / next year. I would call that normal price fluctuation. Last time it was chia. This time AI data centers. Lets see how long people are willing to pay these higher prices.
I think many operators have simply become accustomed to the current payout model and no longer question it. Businesses regularly review pricing and compensation as costs and markets change. Healthy debate isn’t a problem—it’s how better decisions are made. Accepting the status quo without asking questions doesn’t benefit operators or the long-term health of the network.
Well there is a difference. If you believe the general costs of operating a storage nodes have increase (you claimed that) than you need to proof it. You didn’t do that.
The problem I see with that is if node operators cry for a higher payout rate every time they have a chance to do it there will be a time when the higher payour rate might really be needed but the other party is used to crying and ignoring it. So lets be honest with ourself first and only raise the question when it is really needed. That way there will be someone on the other side listening and trying to find a good payout rate. We only have one attempt for that discussion and we should be damm sure we are prepared for it.
I think it’s way simpler than that. very few operators complain about compensation and yet, while they are free to leave the network, they choose to stay. Others don’t complain. Another minority (me) advocate for further reduction, because today’s payout is too much, and still attracts wrong type of operators who are trying to make money on this. Those also happen to be the loudest.
The facts are – there are way too many operators, more are joining every day, further diluting the payouts for everyone. There is absolutely no reason to increase payouts. There are plenty reasons to reduce (or do nothing – and let inflation take care of it)
Long term health of the network will benefit from operators that don’t jump the ship because the payout crossed or did not cross a random threshold of fairness they had in their head.
I’m staying no matter what. If storj pays me $1/month it’s still better tan $0/month, which I would be receiving otherwise. No, I’m not missing hidden expenses like electricity and maintenance. Server are running anyway. extra storj load is negligible in power cost. I don’t maintain anything, it just works. My biggest expense is time waste of converting stupid tokens to real money every month. But this takes me 10 seconds, so $1 is adequate compensation for that. And yet, storj pays me two orders of magnitude more. It’s bonkers.
If in the future storj needs more supply – they can use payout rates to drive that.
Year over year my electricity (/kWh) prices went up by ~24%. The electricity prices per unit are going up each year, but this year was extreme.
Connectivity prices went up by 12% this year for the same service. As they do every year, albeit not to this extent.
Income tax is additional 1 + 6% (+ %%% if you deal in short term crypto). The VAT went up by 3%. And there are talks they might tax us even more the next year. Thank god there is no tax on unrealized returns. For now.
The drive prices on used marked are twice as expensive as they were the last year, the memory sticks went up four fold. The SSDs I won’t even talk about.
I had two drive failures recently - one I would be able to copy the data out, but had no spare. I used to, but currently there is no incentive to.
But of course all these numbers are irrelevant, as we run this on equipment that is already on, but people will scale down and/or not repair. As me.
So cutting the payouts might be risky. People might stop to care. They will harvest the equipment and then leave, as why would they care for pennies.
The network expenses will move from payouts to repairs.
Cutting the payouts will also drive the economies of scale = more centralization on more ghetto equipment.
All this and the network performance will drop. And suddenly you no longer have the fastest, cheapest and globally distributed network.
And the worst case you might loose your data.
But then, I might be wrong. After all, there are people that will donate their time and even money to be able to be part of something.
Yes. You are wrong. Node consumes negligible amount of extra power on top of whst your server woudl have been consuming anyway. So “electricity expensive” argument does not stand.
Nor is “time for maintenance”. Nodes don’t need maintenance. You configure it once and don’t touch them ever again.
I believe you are an extreme outlier on this subject and arrogantrabbit’s opinion covers the majority of sno’s opinions including mine.
If it sounds ridiculous for me to say I’m representing “the majority of SNOs”… well… that’s because it is . We’re only speaking for ourselves… and if others want to chime-in here, they will.
You pay for space to be online anyways: Storj adds pennies. And maintenance consists of watching for “Your Node Has Gone Offline” email from Storj Labs. Easy money.
Some of the people in here do not care, because they have made a business out of this - selling vetted nodes, for example. Like selling shovels during the gold rush.
Or the incentive might be: lower the payouts, SNOs leave, more data for me, because I do it at scale. It is like approaching your employer and asking them to lower your wage because it is too high. No sane person would do that unless there were an incentive for them to do it. And it looks like you might have such an incentive if you are voluntarily asking to lower the payouts.
And this is the centralization I’m talking about.
For me, there are no hard feelings. If this makes me lose money, I’m out. I have done this already with other projects, and year over year it is slowly approaching that point with Storj as well.
Yes, it was my misunderstanding. I upgraded the Internet access during those load tests and expanded the storage when nodes became full. But as already mentioned, I’m now naturally scaling down as the drives are starting to die.
That’s a strech. I’ve seen it proposed a few times, but never actually seen it happen. Care to point me in the direction of where I can bu a fully vetted note?
And with 1.5TB/month, it’s hardly a goldrush. Who would pay money for something you get for free? I don’t get it.
If this makes me lose money, I’m out. I have done this already with other projects, and year over year it is slowly approaching that point with Storj as well.
I just don’t get it. Are you a miner? Do you come at StorJ from a bitcoin angle? If so, you’re doing it wrong. StorJ should be running on equipment you have online anyways. Could be your family NAS that’s hosting your images. You will never loose any money unless you’re running this from an angle you’re not supposed to.
I upgraded the Internet access during those load tests and expanded the storage when nodes became full.
So you did go against the recommendation; don’t pay any money to run StorJ
This contradicts the observations: Nodes count keeps increasing and storage surplus does not reduce.
So even if the majority of sno have different opinion, their actions contradict those alleged opinions.
To reiterate, my point is the for network stability payout shall be small enough to discourage folks with miner mentality to join – they are volatile and cause churn when they leave when it becomes “unprofitable”. Only the very boring operators remain, who are not here to make money, but don’t mind however small offset on expensive of running their servers. This inherently yields much better and stable system.
Have you seen amount of repair traffic lately? It’s insane.
I have no evidence for this, and this is exactly the behavior low payout shall fix.
No, it is not like that at all.
I just described the incentive. Network health is an incentive.
Look, all my nodes pay $100/month. do you really think it makes any dent in anything? If it was $25 or $300 it woudl still not make any difference. But it makes huge difference on the node turnover, as it crosses imaginary thresholds of “profitability” in peoples minds.
This is the issue. This mentality is what I’m talking about. Storj does not want such operators that jump ship when what they think is profitability threshold that only exists in their heads is crossed.
And this is beside the obvious: Node cannot lose your money. It pays money. Without node, you have exactly same expenses, just without storj helping you cover them.
If you disagree, if you built hardware for strorj – then that’s precisely what I’m talking about: Payouts are too high, they incentivize people to bring new systems online to make money. This is bad for the network. Because when common sense hits – they leave, cue repair traffic.
(empathy mode kicks in, did not have coffee yet) yes, it was quite tempting, especially with the storj incessant promises to onboard the new customer and this madness becoming the new normal.
Tangent on the drive dying: start buying used enterprise disks. They don’t die. They are half price. And when they do develop bad sectors occasionally - GHD and similar vendors replace them no questions asked. When I was buying new disks like a peasant I was spending money on mailing RMA almost monthly. Once I switched to refurb/remanufactured/used enterprise disks – 7 years and counting. Not a single failure, outside of minor few bad sectors here and there, which I ignore – 1 sector out of 40 million is absolutely acceptable failure rate. My experience is anecdotal, but not too narrow - I have about 40 disks working total across servers.
Many people keep saying Storj only uses our unused hard drive space. That isn’t the whole picture.
Storage node operators also provide internet bandwidth, electricity, hardware, cooling, maintenance, physical space in their homes or data centers, and their own time to keep everything running reliably.
The network exists because operators contribute all of these resources, not just empty disk space.
Sometimes it feels as if some people are speaking only from Storj’s perspective, as though node operators should simply accept whatever is offered. A healthy network should recognize that this is a partnership where both sides create value.
I believe it is reasonable for node operators to ask whether compensation continues to reflect the real costs of operating the network.
It is certainly reasonable for node operators to ask whether compensation continues to reflect the real costs of operating the network. We definately see the occasional “I’m leaving” post, and that’s probably healthy.
But as arrogantrabbit pointed out… the network keeps growing in node count and capacity. To me, that means it’s reasonable to believe node operators accept their current level of compensation. Is it at least fair to agree more are joining than leaving?
Although SNOs should definately keep watching for other projects that could use their spare space too!
yes, operators do donate bandwidth they likely would not have to have access to otherwise. This rules out home users on metered connections. Organizations (with peering) and home users with unlimited internet does not see this as a problem.
This is being used anyway with or without storj. Server already exists and burns electricity. Hardware already exist. Cooling, maintenance, and physical space is also occupied. Marginal STORJ load does nto require you to buy a new air conditioner. Additional power requirements can be trivially measured and are within few watts noise.
If someone keeps server online, let alone started a new one, just for storj – they are already in the wrong.
One time setup is all it takes. “docker run …” and setup port forwarding. And don’t touch it for decades. What maintenance are you referring to, that is specific to storj, that you dont’ otherwise do for the existing server?
Yes. Excess, already online capacity. It costs you nothing. Therefore any payment over 0 is free money.
True. But storj is unique in the space – it actually converts waste into value. Various chia and other bullshit convert space and CO2 into more CO2 and more bullshit. I would not join a project if I don’t believe in the case, let aline actively against.
On the point – other than storage a candidate is compute. it usually comes with massive additional power requirements, unlike sharing access to storage: While I have plenty of compute available, if I start sharing it my server will start consuming 400 watts instead of 100. This 300W extra watts I did not need to burn, that will cost me $100/month in electricity, let alone defeats the point – someone with more efficient processors may be able to produce that compute power cheaper. This is actively added resource usage and cannot fit the “share what you don’t use” concept, as it works for storage.
The “install it once and forget about it” argument seriously understates what a storage node operator actually provides.
Storj does not merely occupy a few unused sectors on a hard drive. It generates continuous uploads, downloads, disk reads and writes, network traffic, CPU activity and heat. That workload increases power consumption, keeps cooling fans running, adds wear to hard drives and other components, and places a real load on the operator’s internet connection and router.
A node also does not magically maintain itself. Operators regularly check whether it is online, investigate downtime, monitor disk health and available space, install updates, deal with Docker or operating-system problems, replace failed drives, troubleshoot network interruptions and make sure the node remains reachable and reliable. When something goes wrong, Storj does not send a technician to our homes—the operator provides the time, knowledge and labour.
Even “unused” resources are not free resources. Disk capacity has value. Bandwidth has value. Electricity, cooling, hardware lifespan, physical space, maintenance and the operator’s time all have value. The fact that equipment was already owned does not suddenly make its use, depreciation and opportunity cost equal to zero.
By that logic, a taxi company should barely pay a driver because the driver already owned the car, already had insurance and was going to keep the car parked somewhere anyway. That is obviously not how a fair business relationship works.
Storj receives a reliable infrastructure service from thousands of independent operators without having to purchase, host and maintain all that hardware itself. Presenting this relationship as though Storj is simply doing operators a favour by filling their “free space” ignores where a large part of the network’s economic value actually comes from.
Node operators are not asking to be paid for imaginary costs. They are asking people to stop pretending that bandwidth, electricity, cooling, hardware wear, maintenance, availability and human time are free.
. We’re only speaking for ourselves… and if others want to chime-in here, they will.