There is something that never made sense for me:
They want small nodes and they plan to ask for very small downtime (not in place yet)
But IMO only people invested enough to have big nodes can acheive a few h downtime a month. There no way someone will have a few TB node on a simple spare disk and be online 24/7 for multiple years
3 Likes
Mhm, what a bummer… after reading the proposal and most of the comments, I’ve to say that it might make more sense to take my node offline at those new proposed rates.
At 0.52 EUR per kWh, I’m paying in Germany this year, it’s hard to have a setup that is remotely profitable. I know STORJ says to use the stuff you already own, but running STORJ in the first place is/was what made running a machine 24/7 even feasible in my case, and as many pointed out, without a decent return from STORJ one wouldn’t be able to replace any hardware that breaks while running a Node and it would actually be smarter to just sell used drives for an instant payment instead of setting them up to run STORJ to basically earn nothing until the drive eventually breaks.
At the new rates I’d be smart to use the money I’d otherwise spend on electricity & my time to buy STORJ every month - if I wouldn’t start to doubt STORJ’s long-term potential at the same time.
From what I’ve read here I agree with a few things that were mentioned:
-
If STORJ would have an option - as suggested by some here - to say “I want to have 10% of my drive space free” and then automatically scale how much of the space is used for STORJ depending on how much of my other stuff I put on the file it would make those low rates easier to swallow as we could then really run STORJ to just share the space we aren’t using right now but might need in the future.
-
STORJ should charge customers that want to use edge services the extra costs those services cause.
-
Starting to reduce costs at the position that is responsible for 5% according to the information shared here instead of in other areas first seems to be a really dumb decision.
-
STORJ shouldn’t compete just on the price, I see no problem charging a little bit more than competitors.
-
STORJ should definitely focus on getting some certifications and proof compliance with HIPPA, GDPR, CCPA, and whatever else all those fancy regulations are called.
-
STORJ should put some effort into getting a Larvel package and WordPress plugin developed and in general, provide more SDKs for popular programming languages to make it an easier/better choice for developers out there.
-
Maybe STORJ can offer different plans for “backup” and other use cases that are priced differently. The “backup” plan could have a lower $ p. TB/m with a higher cost per TB transferred and other plans could have higher $ p. TB/m with lower costs for bandwidth used.
I’ll have to think about the proposed rates a bit, but it’s likely that I’ll just gracefully exit my node two or three months after the changes are in effect to at least still make some money from the potentially higher repair egress traffic that should be needed once a bunch of nodes starts leaving the network.
2 Likes
that is very cheap, € to $ is 1.06 today in favor of the €, so that takes the price from the current lowest price of 128 to 132$ for the 10x10TB servers
on top of that, these are auctions, so they will keep rising in price until the auction is finished.
on top of that you also need to account for monitoring and other man hours dealing with the storagenodes still staying within the 1.5$ pr TB
if the auction even ends below that mark.
this would also be a central location, so wouldn’t really give the same redundancy as regular random SNOs
but it certainly seems very cheap, and still it cannot compete with the prices Storj has proposed, even tho its in a datacenter which is basically designed to ensure the lowest possible prices.
This is the way.
20 characters
I all,
if these will be the new conditions, I’m out of it. Now I have 8 nodes with 4 different IPs with about 40TB of archived data.
I have been following this project for years, but unfortunately with these new conditions, I really think I will leave.
1 Like
Please correct me if I’m, wrong, but if I want to use STORJ file storage as a e.g. python developer, there are currently two options for me:
- Running my own uplink gateway
- Using the hosted gateway
Isn’t it possible to implement the uplink functionality in a python library? So that I don’t need to run the extra service and instead just configure my python library with my connection details and it will handle the up-/downloads automatically without using the hosted gateway or needing uplink somewhere on my own server?
This becomes even more important if I want to develop a mobile app. It would be really nice to have the uplink as a library for common mobile application frameworks, so the apps can directly connect to the nodes without the need of the hosted gateway or any self-hosted uplink.
I think if that would be available for common programming languages, less people would feel the need to use the hosted gateway, which would lead to significantly lower costs for STORJ.
3 Likes
Aren’t you a bit off topic in this? It’s the edge services that are the costly part which is allowing for link sharing.
1 Like
Yes, but everyone who connects his/her e.g. nextloud or in general S3-compatible service via the hosted gateway is also causing the costs, right?
And for me as a developer, I would just use boto3 and also use the hosted gateway for my app, because that’s the easiest way for me to use STORJ as my backend storage. If that would change and there would be the option to avoid using the hosted gateway by using a STORJ python library, I would be happy to use that instead. This would avoid the traffic costs for the hosted gateway.
And when options like that exist, STORJ could easily charge some extra fees for using the hosted gateway, but users are not forced to use that.
1 Like
Yea I guess your right gateway is what costs but I think that is the edge services there talking about, honestly hearing about edge services is pretty new and I don’t fully understand What exactly it is and what it does and why that has been pushed onto us as if it’s our problem.
1 Like
You’re describing how I’ve run my NAS for decades now… I honestly don’t know why Storj hasn’t had simple NAS installers for nodes as this is the best use case for it. Always on in many cases by default and often plenty of free space. With some adjustments the node could even automatically scale up and down its usage based on free space. It would be completely set it and forget it and most likely very stable nodes. And operators who don’t pay any extra energy bills (or at most 5W due to more activity).
They tried with a QNAP app, but from what I saw, it was a mess and wasn’t kept up to date. Making solid apps for these devices could target the exact right SNO audience.
Not exactly what you’re looking for I think as it still uses uplink-c. But just making sure you’re aware of this option.
Edit: It seems the latest version of uplink-c is v1.7.0 and it’s not being kept up to date… That’s not very encouraging.
It would also make a major difference if uplink was available to run in a browser. I know there are some challenges to that which @jtolio has outlined a while back. But this opens up a lot of use cases to use native implementations instead of the edge services. Getting certs for nodes has become a lot easier over time, so I think this warrants a new look as well.
I think it’s just mentioned as an explanation. Which is why we have the option to redirect that comment into suggestions to begin with. So, all the edge services (s3 gateway, linksharing etc) function as an in between layer between client applications and satellites/storagenodes. Effectively, they translate S3 calls or web calls into uplink functions. So think of it as a cloud hosted HA server that takes requests and then uses uplink to translate them to the Storj network. The problem is that all data flows through them. On upload a customer uploads their data to this server, which then does the encryption and erasure coding and sends the data out to nodes. For this server, sending that data to nodes incurs egress costs that aren’t paid by customers at all. On download, the server downloads data from nodes, recombines the file and decrypts it and sends it back to the client. That transfer to the client has egress costs attached to it again. And all the encryption/decryption and erasure encoding/decoding incurs compute costs.
Now none of this is our problem, hence why I suggested to charge customers separately if they want to use these services. This would take that cost out of the equation for node operators. This inefficiency should not be on node operators to give up payouts for. In all honesty, I think the gatways were initially implemented as sort of a stop gap solution until more native implementations are available. But as with all temporary solutions, it’s looking like it’s becoming quite permanent and I’m not seeing as much movement on native integrations as I would like. I wonder if @john would be willing to share the costs of these edge services so we can get a better picture of how big this impact is. But ultimately, I think it should just be separated out from native traffic and node payments.
8 Likes
here is my calculation.
my first site setup use 450KWh a month- in my prices it is 0.2EURO/KWh it kind of 90Euro for Elektricity.
now Ethernet is 40 euro so we have 130Euro month running cost.
site 2 is around 100KWh it makes 20Euro and 15 Euro internet.
All together is 165Euro is running cost.
today i get around 700$ a month. if it will be only 1/4 of it it is 175$ a month I will not be able to invest to hardware if something go broken. So consideration to end it will be big.
7 Likes
And it’s not just cost. It means centralization which Storj originally wanted to overcome.
1 Like
Yep, it’s server side encrypted and reintroduces all the traditional traffic peering bottlenecks. So no endless scaling using parallelism, not end to end encryption and no built in protecting against internet bottlenecks. Although I should add that these are geographically distributed highly available servers. So all the traditional means to overcome those issues are applied. Emphasis on traditional…
Btw, I can see why native integrations are not prioritized… customers aren’t clamoring for it. But… why would they? They’re not paying for the costs. Storj has taken away their own incentive to build these native integrations and has kicked the can down the road by first swallowing the costs and now trying to make node operators pay for it, removing the incentive to change that entirely, while keeping the inefficiency around. I think it’s time to refocus on the original vision for Storj a little. And make these things happen so the network could work better for all sides.
8 Likes
As i told aboth, this is the place for innovation. Edge service can be decentralize, but for some reason Storj Labs started from wrong End, they announced community satellites and then will be Edge services operators, but if this is the component that makes price bigger, why not start from it and not yet waste spere time community satellites setup.
Edge service can lower price for Storj labs, and rize income for operators, as we dont have metered connections, and have some power machines to run it.
2 Likes
Encryption is a tough nut to crack there. Clients will send and receive unencrypted files from gateways (obviously still on an encrypted connection). You don’t want decentralized nodes dealing with unencrypted data.
And changing the workings to do encryption/decryption locally will negate the drop in replacement nature of the current S3 gateway as it requires more client side changes. At which point you might as well focus on native uplink integration to begin with.
1 Like
I agree that native integration are the best way, but if client prefer not encrypted connection it is his choose. Byt some simple encryption can be done on the fly in brouser or phone.
The auctions on Hetzner work the other way around. They start high and will drop by about 2€ or so when the timer reaches 0. So it is a race in the opposite direction until someone takes the deal.
4 Likes
You do as you wish @BrightSilence. The model you are using for making your business assumptions is extremely simple. The Company is on the market for a few years. I guess the models used there are probably very sophisticated, the number of scenarios and variables significant. Those are the Company’s secrets and there is no way you can obtain this data for additional analysis. For me, the most worrying out of the info provided here so far: i) it seems that the marketing campaign related to free accounts did not bring expected results and ii) I got a feeling the business model is still drifting towards the perfect competition scenario. I really took a look at your spreadsheet a few times, so many thanks for making it public, it is very useful, however, again, I do believe it is simple. Somehow, I like the general idea that stands behind this Company. Even though, I think that the offer is not quite ready yet for a prime time (product still too complicated, communication channels not quite ready, target market not fully defined). However, again, I really do think that the Team is very committed to make it a success and I assess the abilities of the CEO as rather being very high.
1 Like
I will try to summarize my current situation without emotions, with pure facts, just for me and Storj to know, when the project will be feasible for me as SNO. All costs are storj exclusive, simply what I will not pay, if I stop as SNO.
my current setup:
storj capacity 54 TB (30 TB filled)
storj exclusive setup cost 1200 eur (HDDs, some redundancy HW)
expected lifetime 5 years → 1200/(5*12) = 20 eur/mo
current running costs:
electricity 6 eur/mo
extra internet connection speed (faster plan for storj) 6 eur/mo
VPN running cost 14 eur/mo (unlimited data speed and data rate)
total running costs:
46 eur/mo
What is the minimum payout to make it worthwhile to invest my time taking care of this setup long term? Around 50 eur/month for current setup, so minimum feasible payout for me is around 100 dollars for 54 TB setup. By this I mean the payout when it is 50:50 if I will continue or not.
Big thing to consider is, that the setup can not be full long term (fill rate of new HDDs after old ones dies). So I am not able to estimate payout per TB, because important to consider long term is:
- how fast can new node grow
- what will be held amount on new node
- what happens when two new nodes has to be started (two disks dies at same time)
I will wait for announcements regarding Storj staking and /24 subnet limitations, after that I am able to estimate, if it is feasible to continue or not. It is around 50:50 now, I believe.
3 Likes
Ah thanks a lot, I didn’t know this. But seeing that it was last updated 2-3 years ago and it’s a third-party wrapper does not really convince me to prefer this over the hosted gateway. Also the installation is much more complicated. It looks like I need to install C-compiler, the GO-stuff and libuplink…
Seeing this, for a developer it’s much easier to just use the hosted gateway.
And it seems like these (outdated) third-party wrappers only exist for python, php, rust and nodejs. What about all the other popular languages? Especially in the mobile app development I think it would make a lot of sense if STORJ would provide libraries to interact with buckets directly.
Yes exactly. And that should change in order to become profitable.
2 Likes