I think it’s true that at the moment the major issue facing SNOs is the lack of customer data and (by extension) the lack of new data ingress.
Tardigrade finds it difficult to compete. There is S3, the default choice for most enterprise users looking for block storage.
… and, there is Backblaze, dropbox, etc who compete in the home-user/backup space.
I believe Storj can offer something for both markets by offering a lower cost option for home users.
At the moment Tardigrade splits a file into 80 pieces with 30 required to reconstitute the file.
Financially this equates to:
$10/Tb paid by tardigrade customers
$1.5/Tb * 80 /30 = $4 paid to SNOs
$6 profit for the satellite
I suggest the creation of a new satellite Storj Home". This would operate in a slightly different way and would require 50 pieces to reconstitute a file.
$5/Tb paid by tardigrade customers
$1/Tb * 80 /50 = $1.60 paid to SNOs
$3.40 profit for the satellite
SNOs are paid only $1 per Tb instead of $1.50.
The system is less robust of course. This could be mitigated by a few measures:
This server could require 1000 audits instead of the usual 100.
Audits issued more frequently to ensure uptime and issue repairs to other nodes.
Payouts held back for a longer period (e.g. 3 months) for all payouts not just this satellite.
Explicitely require an opt-in for SNOs.
The intention is to make this a good fit for longer term more reliable nodes and make SNOs who wish to provide this service think carefully about their engagement. Everyone will receive less money per Tb, but the increased traffic should more than make up the difference.
The idea of a less robust storage option is interesting, the question for it is how the end user would respond to data loss and how the reduced node count impacts durability.
I’m not sure if paying less to the SNO is a good idea or not? I would personally do a Graceful Exit from the lower payout satellite once my node was full so that I could then continue to collect the more profitable data.
Yes, it would have to be a reduced SLA compared to Tardigrade. The precise SLA would need to be calculated by the Storj team using an average time to failure for a node.
It could be that 50/80 pieces increases the risk of data loss too high. Tardigrade offers 9 9s durability (99.9999999%) with it’s current 30 piece model. Backblaze is currently 11 9s.
However, balanced against this is the speed of ingress and egress that Storj can offer.
Another poster mentioned the chance of a SNO doing a graceful exit when their node is full so they can fill their node with more profitable data. This is expected behaviour.
Backup and user based data would be expected to have a very low egress compared with ‘normal’ storj data, so even if the standard $1.50/Tb is paid the SNO would receive less revenue for backup data. However, if a SNO has a 14Tb HDD running as a node it will be more profitable for them to 90% fill it with ‘Home’ data in 1 month than to wait the 2-3 years for it to fill with Tardigrade data.
If they then choose to graceful exit - there will be others who have space on their HDD to take up the slack.
It makes sens to me as long as there are people with free HDD space willing to accept less profitable data.
I personally wouldn’t mind a lower payout for more ingress however I’m not sure that pricing is the only factor holding people back from using tardigrade.
This would also significantly increase the cost of repair. Since some slack needs to be built in, repair would have to be triggered after only about 15 pieces are lost instead of 28 right now. Almost twice as often. And repair would have to download 50 pieces instead of 29.
Furthermore having to upload and download in more parallel transfers for every segment would also have an impact on network infrastructure, overhead and speed.
Changing these RS settings has a significant impact on many more things than you are accounting for right now. Instead of launching another satellite with worse settings, I would really suggest looking into tweaking the settings of the existing ones to optimize things.
Your suggestion would likely end up not being profitable at all.
Small note… income, not profit. A good chunk of this is used to pay for the satellites and repair services. Then Storj Labs employees need to be paid as well. And that will likely mean that even without taking the massively increased repair costs into account…
This income will no longer be enough.
All this aside. Tardigrade is already a very good deal. The price is not the bottleneck for onboarding customers. More likely this is a matter of PR + feature set. Instead of making the product worse to lower the price, I think the focus should be on expanding the feature set and integrations as well as getting the name out there. And from what I can tell Storj Labs is already working on both of those.
I confess I do not know the precise details of repair, but I think I can comment on some points you make:
Repair costs: You are right. The cost of repair would go up because the repairer would be triggered more frequently and it would download more pieces (although the cost of repair per Tb would remain the same as those 50 pieces would store more data than the 29).
$3.40 income : I do not know if it would be enough, but some of the costs you mention do not scale linearly with traffic (e.g. staff). My hope would be that there would be a margin that would be multiplied by the greatly increased customer-base - yielding sufficient income to pay for both scaled costs and (relatively) fixed costs.
I do not agree that tardigrade is currently a good deal, although in some use-cases it excels. The price limits the Storj network from being offered to anything outside those niche use-cases.
e.g. I cannot:
Backup or store my personal data to the Storj network
Currently Storj does not compete on price.
Use the Storj network as a data store for my business data
Currently Storj lacks the pedigree to be a ‘safe’ choice of storage.
Of course - if the Storj margins are so thin this isn’t possible, then this becomes a moot point
I think 60% comission can be reduced to 20-30% to lower cost of storage.
The commission is very large, and if you consider that the project had an ICO, then there is a double earnings of the company (greed?).
I also think Tardigrade it’s expensive. Of course it’s cheaper than AWS, but everyone it’s. I do my backups on cheap servers with Raid and will prefer to use tardigrade and support the network-idea, but not at the actual price. I don’t have a solution.
I think the multipart upload is a really badly needed feature. What else do you have in mind?
For the integration part I recently started to fear if there is a lack of partner management. If I read something like this:
If an integration partners development is too slow or they even abandon the integration that will get Storj a bad reputation. To me this looks like the partnerships and integrations needs to be more actively managed and eventually even pushed to keep up with the product.
And for getting the name out there. There is indeed much improvement possible. I just had a look at: https://hypernetlabs.io/galileo/tutorials/#tutorials and https://hypernetlabs.io/galileo/tutorials/tutorial-tardigrade/.
Naming alone is confusing. At one point it is named Storj, then it is Tardigrade. We see Tardigrade with Storj logo. Never do we see the Tardigrade logo. A more consistent approach is certainly missing. And of course much more could be done and should be done to get the name out there and make potential users aware of the solutions offered.
The satellite income is already far lower than what storj pays for repairing the data (high egress cost for the servers where repairs are done which is mainly on AWS).
The only fix would be repairs done by the nodes, if I recall correctly it’s something they are working on but it’s a tricky thing to implement and not the highest on their development list.
That’s mainly because storj isn’t trying to compete with “cold” storage services.
I think that the most important thing right now it gaining the trust of the industry, and getting some big clients onboard.
Multipart is coming up. You can already test this on the us2 beta and it works great. As for what else, just log in to AWS or Azure and take your pick of feature they offer which integrate seamlessly with their storage solutions without having to worry about egress costs. Things like accessing data as if it were a database, streaming in real time events to object storage, etc. Now a lot of these will be hard to replicate as storage lives on nodes that can’t do complex compute. But for anything that is possible there should be some similar feature if you want to compete on more use cases.
Let’s take the wyze example. They use Amazon for storage, but their also run machine learning models on the videos they store on AWS to detect whether there is a human in the shot. That example will be very hard to replicate, but I’m sure that’s a reason that ways much heavier on the decision than cost. The bigger the Tardigrade feature set can be, the fewer times these reasons can occur.
Storj Labs pushes out updates every 2 weeks. For FileZilla this is just one integration of the many they have to maintain. Storj Labs can’t exactly force them to align with their update cycle and in the past they’ve shown they will get to these things anyway.
Really can’t argue with you there. Though I think Storj Labs has made it pretty clear that Tardigrade is what the consumer facing service is called. I think they should actually be referring to Tardigrade everywhere as that’s what they are talking about. Partners aren’t under the control of Storj Labs though, so again they can’t really force them. They should still help them clarify though.
They decided to do this in 2 steps. The first step was separating out the repair workers from other satellite services and making them run on remote trusted servers. Hetzner as @kevink just mentioned. This is live and probably enough to no longer spend the entire margin on repair. The second step to bring this process to untrusted nodes is much more complex and much less needed after that switch. So no idea when we can expect that now.
It’s not. I would consider it warm storage though. While it is instantly accessible, it’s not available at the hot storage speeds of the large providers. So it really depends on your use case.
Yes I know. I wanted to test it with Filezilla, but no avail: Invalid macaroon version.
I understand that maybe Filezilla cannot push updates as fast as Storj Labs publishes them.
But this may fire back at Storj Labs and their partner selection process. If I was a customer relying on Filezilla to manage my data and suddenly it stops working because Filezilla does not implement required changes, I would be really upset.
Here comes Theta in my mind again and their idea to offer Edge Computing too. Of course Storj Labs does not have the resources to beat Amazon and their multi billion Dollars. But the nodes don’t have to be low powered. If there is a revenue stream, I bet a lot of node operators would upgrade so that Storj could offer some kind of distributed computing as well.
Maybe to offer just storage is not enough.
I agree. I would even call it Tardigrade S3 in reference to Amazon S3 but as abbreviation of ‘Tardigrade Super Storage Service’.
Backblaze is a much more mature company (13 years) that offers a similar product B2 “S3 compatible hot storage” that is also much more mature (7 years), offers 11 nines in durability, and it is offered at a much lower price (0.005GB/month, $0.01/GB egress or free with Cloudflare). Explain to me if I was transitioning away from AWS, Google, or Azure, why I should choose Storj over Backblaze?
Backblaze also owns their own hardware, so the argument that Storj needs to pay employees is irrelevant. Storj has a much lower operating cost than Backblaze unless Storj pad their C suites with big bonuses.
I find it funny how Tardigrade says “Straight-Forward Pricing” on the front page but doesn’t provide the actual costs for storage and egress but instead just provides an estimated calculator that jumbles the cost of storage and bandwidth together.
I actually have to dig into Tardigrade’s development documents to find the pricing and it is nested way below where a typical person would have trouble finding. https://documentation.tardigrade.io/pricing
$0.01 per gigabyte per month for storage.
$0.045 per gigabyte for egress bandwidth.
$0.0000022 Per Object Per Month.
Maybe, someone needs to redesign the front page that actually displays the “Straight-Forward Pricing”
It’s not that hard to set one value to zero and then the other to figure this out…
I agree about the object fee though, should have a slider or a note for that too.
