Hacker News Reader

Hacker News

Comment by PaulHoule | Hacker News Reader

Full thread

PaulHoule

I think $6 a month for "unlimited data" is going to end badly.

The AWS or Azure price is high, but it's a scalable price, a real price.

View on HN

> I think $6 a month for "unlimited data" is going to end badly.

Backblaze has been offering their 'unlimited data' for over a decade [1], and it hasn't ended badly for them yet.

It's fully sustainable because they only lose money on a few customers (like that one customer storing 430TB for $6/month [2]), while most customers use much less storage than that, so the service remains profitable overall.

The soft limit of its 'unlimited' comes from it being a personal-backup mirroring service and not a fully-external cloud storage, so the service only fits certain use cases.

[1] https://www.backblaze.com/blog/all-in-on-unlimited-backup/

[2] https://www.reddit.com/r/IAmA/comments/b6lbew/were_the_backb...

Backblaze is cool, but their limitations to make $6/month affordable on their end essentially eliminate any large backups. You must have any data locally connected to your PC or it will be purged if not seen within 30 days (so you can connect a bunch of USB drives locally, but this is a hack).

Also their upload speed is capped so you can't just upload at 1GB/s or max out your connection to ingest data into their system, like something you can do with S3.

Glacier Deep Archive is still the cheapest thing really at $1/TB, but the retrieval times and also egress data charges are a big catch.

Yeah, they also don't have a Linux client for the backup product. While you certainly can build a Windows-based storage server, and there's even some interesting storage tech in Windows, most data-hoarders store their data on Linux.

Glacier/Deep is good, but with the 180-day minimum object lifetime, you want to be sure that the data is ready to go into the archive before pushing it there. (You can use tiered storage, but then you're storing all data in standard S3 for 30 days before it gets into Glacier, and that one month of storage in standard S3 will cost you the same as 10-months of Glacier storage.)

I've built a production line for what i call "three-sided cards" that usually have a printed front and back and a supporting web site linked by QR code and/or NFC.

The web site is hosted through AWS S3 and the Cloudfront CDN. The "web side" is carefully optimized for storage and transfer costs. (Glad webp is good-to-go, I am hopeful JPEG XL rolls out fast.) With Glacer/Deep I can afford to archive the Camera RAW, superresolution PNGs and the other assets that go into the print sides just in case I lose my workstation and my storage server.

> Glacier/Deep is good, but with the 180-day minimum object lifetime, you want to be sure that the data is ready to go into the archive before pushing it there.

One of my (to be implemented) backup strategies was to use Glacier Deep Archive as last-resort recovery, and just stick yearly tarballs (e.g. 2018, 2019, 2020) in there. That should save me a bit on retrieval requests as well.

That would be super attractive to me if my ISP didn't have data caps. My personal archive is ~40TB of drives in my primary desktop. Even disregarding empty space and duplicates, it would take four years or more to upload it onto a cloud service without running over my data cap.

I have symmetric gigabit ethernet and have a few TB backed up to deep archive, but if I wanted to backup and restore everything the data transfer pricing is insane, it would be $1800 for 20TB. This really is the only thing AWS is keeping artificially high to facilitate lock-in.

Can you pay to remove the cap? If you can you could pay the extra to remove the cap for however long it'd take to upload everything.

Another option is to drive the data to some place with a fast internet connection and do the upload over the course of a day or two (one time I asked a friend if I could bring my laptop and a hard drive to his university classroom and plug in for an afternoon (I was lucky to know a few of the IT folks there), and I got 940 Mbps upload. It was like heaven!

Only at ten dollars per fifty gigabytes, which is obviously impractical at this size.

It's not nearly as cheap as Glacier, but Backblaze's B2 is $5/TB (and $10/TB to download), putting it between S3 and Glacier. I don't know if the upload speed to B2 is capped; my internet isn't fast enough to tell.

That might be prohibitive if you have a huge NAS to back up, but for moderately large amounts of data (say 5 TBs or less), it seems pretty reasonable.

So considering they say that customer costs more than 2k dollars, that means that the average must be ~1/1.5TB for them to be profitable. That's a lot and not much at the same time.

Backblaze has limitations that doesn’t make it a backup service and since it’s not a sync service either it’s something else. Something that works for some people. People who are fine with all their conditions and gotchas.

For me Backblaze Personal is anything but a backup service.

Yup, I'm currently experimenting with using an AWS bucket alongside [CloudMounter](https://cloudmounter.net) to create an Archive disk for things I don't really need day-to-day, but would like to hang onto.

I'm currently not backing that data up, though. It's not critical, and the chances of AWS losing it are low enough that I'm not too worried. The biggest risk would be myself accidentally deleting it.

One thing I have decided for sure, though, is that archiving lots of small files is awful. Much better to wrap them up in a tarfile, uncompressed.

Amazon Drive is fine. No need to have a mirror of all that data on my computer. With that I could have selected sync or no sync at all. + unlimited photo storage. Amazon Photo and Amazon Drive it’s like a single product.

I've used Amazon Drive back when they offered their unlimited product, and they've lost/corrupted some of my files before.

I personally wouldn't recommend Amazon Drive as a backup destination to anyone.

It's not "unlimited" in a practical sense because as the post said, they're just mirroring the data so you yourself have to have the storage capacity you're asking them for.

There might be a way to trick the client synching the data into believing that it's still on disk by looking at how it checks if the data is still there and writing a filter driver to intercept and modify the result of the call.

Good luck getting a checksum for a partial file that you don't have

Store the checksums? You still need local storage but way less.

True, but perhaps the checksum doesn't need to be computed if the timestamp didn't change. Or perhaps the checksum is always computed in the same way so you can just store the checksum.

It's $7 a month now. They just announced a price rise.

https://www.backblaze.com/blog/subscription-changes-for-comp...

The bandwidth cost makes me cry.

Reply on news.ycombinator.com