I need to store some data I’ll probably never touch, but would be sad if it disappeared. The cloud is the easiest place to shove it, but damn it can be expensive if you don’t do your research.
The way most of these “archive” object storage services work is they store the data on some clapped out cheap HDDs for fuck all money, then if you need to get that data back they move it to the normal/fast object storage where you can download it.
All of that is relatively cheap, but they sting you on the egress/outbound data out of their object storage (i.e: B2, S3, etc) when you want to download your data to your computer. You could put a CDN in front of the object storage, but unless it’s the CDN of the cloud host (i.e: Cloudfront for Amazon, Cloud CDN for Google), you’ll have to pay the exorbitant egress anyways.
I’ve compared a few providers of archive storage to see what they’d cost. I’m using 15TB of data as an example as that’s roughly how much I need to keep. All prices are in USD unless otherwise stated
Amazon S3 Glacier Deep Archive (Ohio)
Storage: $0.00099/GB/month = $14.85/m
Restore: $0.0025/GB = $37.50
Egress: $0.09/GB = $1,350
Amazon S3 Glacier Deep Archive (Sydney)
Storage: $0.002/GB/month = $30
Restore: $0.005/GB = $75
Egress: $0.114/GB = $1,710
Google Cloud Archive Storage
Storage: $0.0012/GB/month = $18/m
Restore: $0.05/GB = $750
Egress: $0.18/GB = $2,700
Microsoft Azure Blob Storage Archive (West US 2)
Storage: $0.00099/GB/month = $14.85
Restore: $0.02/GB = $300
Egress: $0.08/GB (first 100GB free) = $1,192
IBM Cloud Object Storage Cold Vault (sjc04)
Storage: $0.006/GB/month = $90
Restore: $0.05/GB = $750
Egress: $0.09/GB = $1,350
Oracle Cloud Infrastructure Archive Storage
Storage: A$0.00339/GB/month = A$50.85/m
Restore: free
Egress: $0.0085/GB (first 10TB free) = $42.50
Scaleway Glacier
Storage: EUR 0.002/GB/month, first 75GB free - EUR 29.85/m
Restore: free
Egress: EUR 0.01/GB, first 75GB free - EUR 149.25
OVHcloud Cloud Archive (Vint Hill)
Storage: $0.000003/GB/hour (730 hours a month, so $0.00219/GB/month) - $32.85
Restore: free
Incoming data: $0.11/GB = $1,650
Egress: $0.11/GB = $1,650
Backblaze B2
Storage: $0.005/GB/month = $75/m
Restore: not needed, B2 is “hot” storage
Egress: $0.01/GB = $150
Wasabi
Storage: $0.0059/GB/month = $88.50/m
Restore: not needed, Wasabi is “hot” storage
Egress: free
Dropbox for Teams Advanced
Storage: A$363/year or A$36.30/m for “as much storage as you need”
Restore: not needed
Egress: free
Backblaze Personal Backup
Storage: unlimited data for US$130 for 2 years
Restore: free
Egress: free
There’s two clear winners here - Oracle and Scaleway.
EUR 29.85/m (approx A$47/m) from Scaleway vs A$50/m from Oracle to store it, but if I need to restore the data, Oracle is much cheaper. I’d probably just go with Oracle to be honest. Never thought I’d say that, but here we are.
Notes
Amazon S3 Glacier Deep Archive
Amazon can send a HDD of data to you instead of downloading it (Snowcone and Snowball) - unfortunately they won’t send data outside the country it is stored. So if I keep my data in Ohio, they won’t post a HDD to Australia. Snowcone is limited to 8TB and isn’t available in the Sydney region, so my only option besides downloading the data (for $1710, lol) is a Snowball, which can hold 80TB. It’s US$300 to send it out and they charge $0.04/GB to load up the data. For a 15TB restore I’d be up for US$900 - almost half the price of downloading it. But I’ll be paying twice as much per month to store my data in Sydney instead of Ohio…
Google Cloud
data must be stored for a minimum of 365 days! “You can delete, replace, or move an object before it has been stored for the minimum duration, but at the time you delete, replace, or move the object, you are charged as if the object was stored for the minimum duration” - this locks you in to Google unfairly IMHO.
Backblaze B2
Backblaze can send a “Fireball”, a 96TB Synology NAS with my data on it for a $3000 security deposit that’s returned when the unit is returned within 30 days. Shipping there and back is likely to cost more than just downloading the data at US$150. They have an 8TB HDD option, but even sending that back and forth isn’t much cheaper than downloading the data.
Dropbox
Will let me store heaps of data for relatively cheap ($36/m) without any restore costs, but you need a minimum of three people. I could probably find two others to join my “team” very easily. The other problem is that Dropbox needs the files stored somewhere and active, so I’d need a PC/NAS running Dropbox logged in and with all the data stored locally.
Backblaze Personal Backup
If you don’t have the files “active” on your storage medium for more than 30 days, they’re deleted forever. I can pay $2/m extra and keep it for a year. I can pay $0.005/GB and keep it forever, which is the same as their B2 product. So in practice, this means I have to buy 15TB of storage and shove it in a computer that is on at least once a year. It poses issues when my storage grows as I need to keep a local copy - an expensive proposition once you hit 20TB+.