Yeah, let’s stick to the “hardware I already have” idea. Well, I have a file server with a few TB free, but it is slowly filling up with my own files. So, I either dedicate the space to Storj (in that case, I could have just used smaller/fewer drives) or not, but I cannot let Storj take the remaining space for the node and then release 20GB or whatever at a time when I need to store my files.
1 Like
You could easily do that with NAS systems though. That’s how I started. I never made any purchase in the beginning. Had 10TB spare and figured I’d give this a try. The question is, is there enough of that kind of spare space to serve these needs? Hard to say.
Node operators who optimize and shape traffic for more profit. Keep in mind, the code is open source, you could shape the code to do what is most profitable. If incentives for some type of egress drops, just slow that stuff down or don’t bother with it at all to make room for more profitable work. There’s a lot of trickery you could do to just decline the least profitable data/transfers and only accept what’s most profitable. You want an incentive system that pays fairly so that trickery won’t have any advantages.
Same as above. If you don’t pay for egress, why even serve data. I’ll just only respond to audits and save my bandwidth.
It would be really awesome if you could let Storj just use the HDD up until it’s 90% full and automatically scale down and transfer data off your node when your own use goes up. No need to think about it anymore. It would kind of be set it and forget it on always on hardware.
5 Likes
Just wanted to let you know how I look at this as a Storj user, not a node operator. This to me looks like an ordinary cash grab that puts the whole network at risk and thus causing me to move my data away from Storj.
I’m using Storj for my backups having used Backblaze before. I’m attracted to the concept of redundancy and decentralization. However there must be enough motivation for node operators to operate a node. The hardware in use does cost money and energy prices have skyrocketed. At least here in Europe. If too many nodes now drop out because it’s no longer economically viable to run a node 1. my data is lost if it’s not first transfered to other nodes and 2. it’s no longer decentralized so why bother with Storj at all?
Storjs is less expensive than e.g. Backblaze but not by much especially when there’s not a lot of egress which is the case with backups. In this proposal I’m not paying less for storage and data transfer. There’s just way less money going to node operators and more money for the organisation behind Storj. However with Backblaze I at least know that my data is safe which it no longer is with Storj.
I hope you reconsider but I think the damage is already done and question the long term viability of this project.
17 Likes
At current state they are not gaining profit. They need to either cut SNOs payments or raise price for customers [edit]to start gaining anything when network scales up[/edit]. Otherwise they will keep taking money out of their pocket, not putting it there.
Didn’t know it’s possible to just run however changed version of a node. Well, in that case, if people are just going to do so, having to pay for repairs is necessary…
1 Like
It seems that Storj team already made up their mind, and whatever we request here, it’s useless, it is not a negociation. But for the sake of discusion, here it is…
The truth is that we can’t give a good answear of what should be the prices; only Storj team can, because they have all the data, and when they make a final decition, I hope that they take into account these things:
- energy cost per TB (it’s available on the internet from allover the world; Storj team knows where the nodes are within an acceptable error margin). An HDD consumes aprox 6-7Wh + CPU cycles + RAM cycles + chipset cycles + UPS and PSU efficiency, that gives like 10-15W per HDD. So let’s take 10W per HDD. This will make smaller HDDs under 4 TB unproductive with a price of 0.25€/KW for ex.
- the number of nodes that will shut down after the payout change (I think all nodes smaller than 3-4TB allocated).
- can the remaining nodes process all that repair and take all that data from the closed nodes?
- how bad closing all those nodes will affect the transfer speeds and the delay, beacause now you are putting out some great numbers, but when the network shrinks, the speed will be affected also.
-Changing the payout scheem dosen’t limit just creations of new nodes, witch I’m ok with; but also closes old nodes. This should be done slowly, so the payouts should also drop slowly. - how much the traffic for the remaining nodes will increase to compensate for the payout reduction, to keep SNOs motivated.
My nodes are all NAS with 1 disk for now; they consume 15W per node, so 3€ per month in energy. I will stay motivated to run them if they make at least 20€ per node per month after first year.
My argument is, there’s no actual cost incurred to me when traffic is received or sent through my internet connection. What costs me is electricity and hard drive space. Simplified I need an internet connection whether i run a node or not. I don’t need to run a node and incur the cost of electricity or the extra TB of hard drive space I’ve got as a domestic user. So as long as Storj paid me enough to cover some power and if i choose not to spend payment on other things and eventually buy another drive I’m happy. Again I’m speaking for myself. Surely if you wanted to game the system and say just hold data to not then serve it again, that would be an easy fix?
2 Likes
You had an advanage over everyone else though since you had the storage already and plus starting when V3 started so you got to take advanage of the surge payouts. Which allowed you to be able to pay for more hardware, As it stand right at this moment and the future this isnt gonna be possible to do. That and the fact that most avg users arent gonna have a bunch of un used space just sitting around on there NAS to create a bunch of IO that would probably put there real data in danger because it could kill there hard drives faster.
2 Likes
Phrased it wrongly. Changing it now won’t suddenly make everything positive, but will allow so when network scales. If scaling happened now, the -0.5 would become more like -50. At least that’s how I understand the situation.
Keep in mind this is a proposal and not an announcement. It is phrased that way for a reason.
This is the kind of data that is valuable for the team. I would love to see more calculations like that. I did it for my own storage node but I am not sure how representive that is. So please share your calculations with us.
There is no fixed date. The plan is to reduce the payout slowly over time. So each month a few storage nodes will exit the network. They don’t leave all at once.
If we just stretch the time long enough that shouldn’t be a problem.
Why do you come to that conclusion?
2 Likes
It is “tokens for the repurchase of company shares and other payments including general operations and liquidity purposes not otherwise described above.”
You can read more is the STORJ Token Balances and Flows Reports. For example the last one Q4 2022
4 Likes
@jtolio in one of the posts told that Storj will allow run Edge servises on node operators side.
But ithink there will be some agreements also.
My counter argument would be, I could rent a cheap VPS server and avoid paying egress costs by blocking all outgoing traffic. Cache in on storage alone, while never ever serving any data to customers. The point is, not paying for egress creates a skewed incentive to not bother about egress. Which is not a good idea.
7 Likes
For me there are no reasons to run nodes with new price models. Even with $5.00 per Tb for egress / repair and $1.00 for storing, i will quit next stop.
1 Like
So to summarize the current situation:
- There is not enough demand for storage nodes to be filled
- S3 performance crashed under load
- S3 is expensive for STORJ and NOT decentralised
- Nodes make up for only 5% of the costs. That is somehow to much!
This brings me to the following questions
- Is the recent data increase mostly due to artificial test data?
- Is S3 without a gateway even possible? If yes, ETA?
5 Likes
Not sure if that helps:
- Icy Box IB-3740-C31
- 2x WD Red PLUS NAS, 10 TB, HDD, 3.5"
- RPi 4B
Average load of 32 Wh = 0.768 kWh per day x 0.40 EUR energy cost = 0.3072 EUR per day * 30 = 9.216 EUR energy cost for 2 HDDs + 2 operating nodes per month.
If the HDDs give up their lifes after 2 years, I would need an income of constantly 33 EUR per month, which is not the case at all already. If they last 3 years, it would be 25 EUR per month, which is also not the case for now.
If the earnings are dropping compared to the earnings today, it would be wiser to invest the money into a monthly savings plan instead of burning it. Meaning: we’re talking about a lot of money invested over time from SNO perspective, beginning with energy prices and replacing damaged hardware.
1 Like
End of February I started my first node.
Initial thought was: utilize some free disk space on my ZFS NAS which is always on anyhow. It was a more altruistic motivation until I realized that I would shoot in my own foot. Over time my ZFS pool would become slower and slower (already at 50% disk utilization).
Second thought, add (as recommended) a used drive as a seperate pool. No redundancy. OK extra wattage and I can loose everything.
Third thought: why not to turn down my NAS (keep it for backup of my own data) and operate my DL380 GEN9 with 8 SFF bays in 24x7 mode. Our own (hot-)data would be on 2 ZFS-mirrored SSDs (from the NAS) and the remaining 6 bays would operate 6 storjnodes on used 1TB WD RED PLUS 2.5 (<2Watt each, 6 single stripe vdevs pools). In case a node fails it is just 1/6th of the risk and this would provide full capacity (5,5 TB), no filewalker and IO issues). The increase in electricity compared to the NAS would be around 20-25 Watt.
Now having read this proposal I have a 4th thought. It doesn’t make a lot of sense to operate storj node. The proposed new pricing (and it might not be the last adjustment while I have sunk cost) would not justify the additional maintenance/supervision and cost. The watt per Terrabyte is what counts at the end. And may be it is more lucrative to sell the used equipment before it has no value.
So I assume apart from some enthusiasts who like to contribute altruistic we will see with declining compensation (as proposed above) a shift to larger nodes while smaller nodes should disappear. Isn’t this the rule for economies of scale.
Overall I get the impression that the end customer pricing seems to be the issue since some of the comments clearly state that smaller SNOs are subsidizing the STORJ business model.
3 Likes
I don’t know how they would do it, but giving an encryption key to an anonymous SNO makes the SNO capable of reading customers data directly is not possible, as that will probably breach the base of the Storj customers trust in the company.
I know (probably) you and me will not take a look at the customers data, because I personally don’t care, but just the possibility of a bad SNO filtering customer data makes that service offering hard to believe.
1 Like
Just running on unused hardware won’t work I think or at least not how I would handle it. If I don’t see a need for more storage for myself I wouldn’t have anything unused to begin with. If I think I would need more in the future I would just buy:
- as much I think I will need or
- just a bit more to save on $/TB with bigger drives.
So with the new HDD you have space unused but it will be less over time as I need it till it’s “zero” again and time to buy more. In both scenarios I wouldn’t have a fixed amonut and have to resize regularly. It probably even had time get used much. It’s a constant up and down on capacity.
The lower the payout the more you have to scale in order to justify the time end effort to keep it all running. Even the upper limit may be critical to 1 or 2 HDD providers of “unused” storage.
For the lower limit I’d just put some money in an ETF, savings account, with rising interests these days or something like that and just buy the storage I need from those earnings. So no “unused” storage is available and you can forget about the responsibility as you don’t have to do anything or worry about a node.
3 Likes
I agree with what your saying to game the system, but i also said surely this would be an easy software fix or done through audits that would check nodes to only pay them for distributing data when required. I reiterate what i said the cost is the storage not the connection.