"There was a post about it from BackBaze. They know some people abuse it and put several Tbs on there, but they still make money on average as most people have 100s of Mb ..."

... and this is a mistake.

The issue is not whether you will tolerate loss-leaders or not. The issue is that you create an antagonistic relationship between provider and consumer.

It doesn't matter how nice you are (and the backblaze folks seem very, very nice) - this is a structural reality that will produce predictable bad outcomes.

Maybe not today and maybe not tomorrow, but if your organization creates opposing interests - antagonistic relationships between the consumer and the provider - someone is going to lose:

"... paying a flat rate for unlimited storage, or transfer, pits you against your provider in an antagonistic relationship. This is not the kind of relationship you want to have with someone providing critical functions."[1]

[1] https://blog.kozubik.com/john_kozubik/2009/11/flat-rate-stor...

The relationship is only antagonistic if the provider isn't willing to actually commit to the product they're selling.

If you sell unlimited, you're selling unlimited. It's not an antagonistic relationship. It's only antagonistic when they decide that "Unlimited" needs some asterisks behind it.

The business of selling truly unlimited might not be feasible or viable, but this is more a transparency issue than anything. If you're not willing to actually provide unlimited, don't say that you are. It's that simple.

It's hard for me to feel sympathetic to Google/Amazon for having to store someones 1.8 Petabyte porn collection because that's what they sold them. Clearly they've decided that selling unlimited doesn't make business sense, and that's okay. They're probably right!

But if you offer a product, you should be delivering that product. Anything less is gross.

> If you sell unlimited, you're selling unlimited. It's not an antagonistic relationship. It's only antagonistic when they decide that "Unlimited" needs some asterisks behind it.

If the business sells unlimited because it does not want the general consumer to feel constricted with online storage, then it is a benefit for the general consumer. When abusers of business products test the business' abilities, they ruin the business' chance of offering a great product that the consumer can find convenience with.

All-you-can eat buffet should be treated as such. But when a bunch of big college guys decide to hang out at an all-you-can eat from 8 am to 8 pm studying and eating, it defeats the product. It defeats the offering for others. And it takes away the experience of going to a place never having to worry about ordering more food than you can eat.

This ain't about feeling sorry for the business. This is about feeling sorry for the consumer. It is one abuser that hurts consumer. And that is bad economics for all people.

I'm somewhat more sympathetic to this position in this case as Google was truly providing Unlimited (until now), essentially.

My issue is more that many things are labelled "Unlimited" but they aren't. They aren't even close. Cell Data plans are Unlimited* (Up to 15GB tethered, Up to 25GB at 4G speed, Video limited to 720P)

The thrust of my point is that if you say Unlimited, you should mean it. If that means nobody says it, I'm okay with that. They have a business to run and they should do what they need to to make it work.

But I don't think they should be able to get away with selling it as "Unlimited" when there are several fine print details to define what that means that are as hidden as possible so that you don't know what they are until you've been parted with your money and run up against the limits.

The amount of people who will stay from 8 am to 8pm will be small and should be taken into account when deciding to offer something without limits.

You cannot have the cake(advertisement of unlimited offer) and eat the cake(limit the unlimited).

> The relationship is only antagonistic if the provider isn't willing to actually commit to the product they're selling.

I don’t think the capacity exists to continue serving “unlimited” storage long-term. Over time, the amount of free storage people use increases until you can’t sustain the practice. It’s just a matter of time.

The good bed is that paid storage is ridiculously cheap.

I don't think that's rsync's point. The point is that when someone offers you unlimited storage for a flat rate, your incentives aren't aligned, no matter how good the other party's intentions are. I recommend reading the linked article. It's short and succinct.

It depends on what drives value. Backblaze’s value proposition isn’t bits. It’s just a different business model than yours — and I think rsync.net is awesome for different reasons.

Backblaze kind of limits this by requiring their backup client to have seen a given file in the last 30 days. Beyond that the file is eligible to be pruned. So in practice people dumping data into Backblaze can only keep it there for 30 days, or must have an equivalent amount of storage space locally.

And they deliberately don't provide a Linux client, to deter people from backing up the entire contents of their NAS.

Still some people found ways to trick the macOS client with remote block storage (iSCSI, ATA over Ethernet, USB target mode).

Incorrect. For an additional $2/month Backblaze will keep one year of version history.

They also have a "Forever history" offering, which is essentially a plug-and-play integration with B2, priced at $2/month + $.005/GB/month for versions deleted more than 1 year ago.

Sure, you can up the window to a year, or pay the normal rate for B2 on top of your subscription (albeit you don't pay transfer costs).

However, you still need to rotate the data and Backblaze doesn't exactly make that process easy for this use-case. You need to go into the UI and manually select the data you want, wait for it to be packaged up, and then download and restore them to the proper locations. If the files were deleted you also need to mess around with the timeframe in the UI.

This sounds like a poor UI, but for their normal backup/restore users it's fine since you generally want to restore an entire device or you're picking out a particular file you need. This really only breaks down if the size of the device backup exceeds the actual storage capacity of the device, which is probably a rare-ish occurrence. But also, restoring from backup is something that should rarely have to be done anyway.

My point still stands: Backblaze is able to offer "unlimited" storage by making the backup/restore use-case easy, but making the "additional cloud storage" use-case tedious. This generally puts an upper bound on that "unlimited" amount based on how much local storage users actually have.

30 days vs. 1 year doesn't make a big difference unless you're hardcore about cycling data around as if kiting credit cards.

And the files dumped off to B2 are basically irrelevant to the "unlimited backup for $6" portion of the product.

Aye, but the problem is the PBs, not the TBs, from what I can tell.

One thing nobody really says is that backblaze actually requires everything you've backed up to be accessible to the computer at least once a month, so you can't actually delete something you've backed up unless you're willing to lose it in a month, or pay extra for the 1 year retention tier. Essentially you need to own the physical drive space to store anything on backblaze, so there's not much reason to abuse it.

Wonder how difficult it would be to patch the client or just make some sort of fake FUSE filesystem to bypass this.

IIRC their client detects FUSE and friends. People have used iSCSI or ATAoE to bypass this, but that comes with it's own limitations. Most people don't bother.

I doubt $2/month is a disincentive to a hoarder whose only alternative would be to pay per GB for the storage space they are using.

That only allows them to delete things for less than a year, they would still have to develop some sort of system for redownloading all their data, putting it in a backed up drive, and leaving that drive plugged in for about 2 days for backblaze to actually notice it

Backblaze charge for egress so it makes no sense to abuse it with constant 500GiB/hour of urandom.txt or porn or Blu-ray rips or TV recordings. It’s only useful/usable for backups(as should be!)

The post was about the Backblaze backup service, not B2. You can't abuse B2 by putting too much data on it!

And B2 is more or less fine for streaming. The bandwidth bill will be significantly less than the storage bill.

Egress is more expensive than storage: 1c/GB vs 0.5c/GB-month (https://www.backblaze.com/b2/cloud-storage-pricing.html).

Those are different units. You're making an assumption that the user watches every video in their library once per month. I doubt it. My bet is that someone with a big video library will watch significantly less than half the content each month (including repeats).

Or in other words, look at it more like 1c/GB vs. 6c/GB-year.

I thought you can combine B2 and CloudFlare to get egress for free?

This is literally the Dropbox business model—at least per the Harvard Business School case!