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.)
> 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.
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. (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.)