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

So functionally Storj is the same/similar, but its ace is built-in geo-redundancy?

So Storj should have price a bit higher, somewhere betwen 1-1.5x of Backblaze, clearly stating that if you want geo-redundancy there, you need to basically pay 2x. If there are other benefits, the price should go even higher then.

That’s my understanding of it more or less. I’m sure others can provide more specifics. And although they are relatively cheap compared to other tech giants, the company is making the money with their own equipment and not acting as a middleman so they can work with smaller margins. They have a very unique setup which I think is really good and quite honestly hard to compete with in terms of pricing. But you don’t see everyone else rushing to slash their prices to compete with Backblaze either do ya?

Storj is better due to the inherent decentralized traffic and use of parallel transfers. Backblaze relies on centralized data transfers from a single point (unless customers choose to manually distribute and pay for different region storage, at which point it’s a few centralized spots instead of one). Generally the storage giants are faster than price competitors, but Storj can outperform both in large file throughput.

Generally speaking the storage giants are better at this than price competitors, but both are better than Storj for now. This is because Stork requires negotiating with a satellite before knowing where data is stored on which nodes or knowing which nodes to upload to. This requires an additional data exchange before transfers to and from nodes can start. Storj is working on several initiatives to improve this, but at the moment they are probably worse than most traditional storage solutions when it comes to time to first byte.

The backup cases are mostly cold, but database backups while not frequently downloaded often require short recovery times when those backups are needed. So traditional cold storage would be too slow to retrieve that data again. In those cases Storj can provide hot storage recovery times for cold storage prices (or close to it).

Btw… Backblaze is burning through their runway faster and faster as they grow. They are currently running at a loss as well. This is not abnormal for a startup, but it shows that their prices probably aren’t long term sustainable either. Backblaze, Inc. (BLZE) - Net Income (Annual)

2 Likes

I was not aware they were at that much of a loss. But for them it’s not as bad because they have assets. Storj does not. Backblaze will take the loss now in order to grow and reclaim that loss later on. Storj doesn’t have enough tokens to sustain this long enough to make it to that point especially at the rate they seem to be going through them… and as of Q4 of last year, sno payments only account for about 4% of what they went through assuming I read the report correctly. So reducing that a little isn’t exactly a game changer.

It is if they continue the growth trajectory they are on. If they get profitable unit economics going they’ll be a step ahead of backblaze. But yeah, they need to keep accelerating, hence why they probably don’t want to raise prices right now.

Amazon Backblaze Storj
Latency 10 9 5
Large files throughput 9 8 10
Small files throughput 8 8 2
General parallelism 8 7 10
Geo-redundancy 3 1 10
Encryption/ Privacy 5 5 10
43 38 47

Trying to come up with a sort of comparison then. Tried to pick the winner in each and put others in relation to it. Small files are the exception, as I believe there is always room to improve on transfer of small files :stuck_out_tongue:
I omitted the price cause it would require more number-crunching than I am willing to do. It’s somewhat embedded in the geo-redundancy though. Redundancy means caching as well, but let’s assume it’s embedded there too.

General parallelism - for Storj it’s obvious; for others I meant that they are able to deliver multiple files in parallel through their fast and multiple fibers

Geo-redundancy - Storj has built-in; they have, but at a cost. Though I think that Amazon is not charging 100% more, as Backblaze does, for each copy and is automated, right?

Encryption/Privacy - this is clearly from privacy perspective, not just safety; if notes in my head are correct, I think that Amazon or Backblaze could still decrypt your data if government asked politely, right? In Storj this should not be possible, though what about the uploads via Gateway MT?

How far from reality my values are? :smiley:

Why even this comparison? So I know what’s the place of Storj on the market, what’s the growth potential based on it, what obvious dangers there are. This plus the price for SNOs would let me decide if I see myself as an Operator in the future.

1 Like

Yes, but currently even if Storj reduces payouts down to break even they’re still not making anything. Amy more than that and I would imagine snos will just quit. To potential customers this is a clear indication that pricing will eventually change, meaning it will go up or the company will sink. There’s a whole lot of uncertainty there as it’s not sustainable. How would this be reasuring enough to large customers?

Also, Backblaze caters to the consumer and small business markets as well which is where this pricing really get’s the attention. They even have an app for mobile devices to access your data. Correct me if I’m wrong, but I believe this approach was how they really got their foot in the door. Storj however has no interest in the consumer market and wants to immediately go after the big fish which in many areas of business generally isn’t the most effective approach to start with.

I think you got the winners correctly, but the 10 point scale would lead to a lot of debate, haha. I don’t like handing out 10s as it implies it’s perfect and everything can be improved. I would also not hand out as low scores as you did for any of the providers. I’ll add some notes on your comments.

Latency I think I could agree with for the most part, but I’d drop Amazon to 9 and backblaze to 7. I think Amazon likely has a bit bigger of an advantage.

Large file throughput, let’s get rid of that 10 and drop em all 1 point. :rofl:

Small file throughput is an issue with Storj when retrieving single files or without parallelism. However, when transferring lots of large files, parallelism can overcome most of the issues for Storj. I’d put them at least at a 5 as it can be resolved for some use cases but remains a weak point. My guess is that Amazon will also outperform Backblaze here.

No argument there. But I don’t like that 10. Storj will improve even more with network growth and could improve things by tuning RS settings. I also think that uplink should probably transfer in parallel by default without requiring additional parameters. People often think Stork isn’t that fast because they use default settings.

Both Amazon and Backblaze do have options for this at a cost, I’d give them probably a 5 and a 3. Maybe even higher. And let’s give Storj an incentive to improve by dropping it to a 9 :wink:

Governments don’t ask nicely, they tell and give you a gag order along with it, haha. Yet I don’t think it’s fair to give anyone a failing grade here. Storj does enforce encryption and make it a lot harder (basically impossible) to accidentally have your bucket be public. But security is solid on all platforms. Any issues on Amazon and Backblaze are usually the result of user error. Amazon is likely a step ahead of Backblaze here too.

I don’t think it’s reasonable to think they will stop there. I would expect them to set payouts to start making a profit on unit economics. But lets wait until they publish their suggestions.

I would say they only dabble in that. And I doubt this has any big impact on their revenue. There is not a lot of money to make in that segment, because volumes are always low. The upside that this brings is that more people will know about it though. Having some consumer solutions can help raise brand awareness. But Storj has made some steps in that as well, for example by partnering with ix systems and being the default cloud sync provider on truenas.

This is the main reason I mention it. Obviously the big customers are where the real money is, but It helps establish your brand, prove your product and help you grow while your still small. Consumers are far more apt to jump onboard just because the price is right, whereas large companies are going to be doing their homework before making any risky decisions. And from my perspective I would (and do) use Storj for important things I want to back up (meaning I have my own copy), but I would not use it (yet) if I had PB’s upon PB’s of critical data, mainly because I can’t see how the company makes any money or even intends to in the future at these rates.

1 Like

These are not absolute values. As I wrote:

So it’s relative and “currently”. If I were to assess the future, I would make everything into 1s, as everything will eventually move to 25GbE, NVMe, etc., and then it will move to something even more robust, making any comparisons not possible :V

So 10s are “best in category now”. Though could change into something like “8 (10)” as “it’s 8 now, but will be 10 when at full potential” (that we know of already).

But the cost is basically “buy the service second time” for Backblaze and "yeah, we can replicate for XX% more) for Amazon. And it’s only ONE copy. So if you want five, you go 5x the price with Backblaze (I know they don’t have that many DCs currently) and probably above 2x with Amazon.

Hmm, for Satellites. And possibly in general for some sort of distribution algo improvement, though by principle it’s already there.

I know. Could add a winky face though :wink:

No. This is specifically about privacy. Could move out Encryption into separate row, but Amazon would have 10, BB and Storj 9, so a pointless metric.

No, there is no copies, we uses erasure coding, see

Sorry, copy isn’t the right term. Just being lazy referring to the expansion rate at 2.76 so almost 3x the data stored.

Storj uplink uses encryption by default and the data should be safe from snooping if you do not give the key to anyone. However, if you use other Storj services (gateway, linkshare) you have to give the key to them, so in that case Storj may access your data.

However, when using any service (Amazon, Backblaze, Storj or a public FTP server) you can always encrypt the data yourself before you upload it, this way the service providers will definitely not be able to access it without somehow breaking the encryption.

That couldn’t be further from the truth. I love to learn something new. I love to change my mind about something. And I have no problem admitting that I was wrong. But I need to see something new to change my mind. You know, something like a real testimonial or benchmark I asked for multiple times.

You think that I disagree with you on a lot of things. That is not true. I just don’t simply believe your claims without seeing any evidence. You totally could be right about all this stuff!

Because STORJ are sane people and not crazy.

We can pull numbers out of thin air all day long. That is a lot of fun, but not how the real world works.

HIPAA: That alone can be a no go

Georedudancy: This one is most of the time about performance and not HA. If your business wants that kind of HA, it probably won’t choose a startup S3 gateway.

Peering: This is a lot of time underestimated when it comes to performance. An AWS VPS will have good peering to an AWS S3 bucket and the same is true for Azure or any other cloud provider.

Features: S3 feature complete? Versioning up and running?

As far as performance goes, I will spin up some VMs and run some benchmarks in the next few days. Maybe that will change my mind :wink:

Did you look at the case studies I linked? You don’t have to answer that… there’s no counter next to the link, so I know nobody has clicked it…
Doesn’t really incentivize me to look for more links for you. Sorry.

2 Likes

I don’t need to click on a link to see that you only linked to the default testimonials page. That’s what hover is for. That page is full of weak testimonials. I walked trough that mud once. Was not worth it.

So no, I did not click on your link that points to a page I already criticized previously in this thread :grin:

So basically I’m going to link my wallet address here, if anyone could chip in with something, take a bit of loss please… but hey at least I can save costs, I can afford more things…

Its a nice idea, but I dont think, personally, will give more so companies can turn a bigger profit. Id rather idle the drives for myself. Sorry but not sorry.

I did a very basic benchmark. I choose a scenario, where STORJ should excel:
Uploading and downloading a single big 4GB file (ubuntu iso) cli.

Average Upload STORJ: 13.21MB/s
Average Upload Backblaze: 103MB/s
Average Download STORJ: 17.8MB/s
Average Download Backblaze: 70.9MB/s*

That is why I don’t think that STORJ competes with AWS performance.

  • Not sure if Backblaze, SSD write speed or download_cap_exceeded
    slowed down this one
1 Like

I think your setup was wrong. Did you used parallelism parameter.
when i tester myself my 500mbit convection screamed on booth way upload and download was max speed of my connection.
Besness made it even on 10Gb connection with max speed.

8 Likes