There was a length discussion over on HN about that. tl;dr: Write once, read never.
I’d want to see some official Amazon cost calculator to understand the cost of restoring one of my 2T drives.
It almost seems like the only use case for Glacier is for legal CYA reasons, where an astronomical cost of restoration still beats the legal costs of not having the archive to being with.
I imagine there might be other use cases—for things where you expect a real-time response, as in when a user is waiting on data, it clearly isn’t valuable, but programs are more patient. If you have many asynchronous tasks, you can get a lot of work done very cheaply with this, especially if for the more common cases you hit a cache which can resolve more quickly, and only for the less common ones you hit glacier.
Dirt cheap, wish it wasn’t so, ahem, glacially slow.
Wasn’t there a concern about retrieval costs?
There was a length discussion over on HN about that. tl;dr: Write once, read never.
I’d want to see some official Amazon cost calculator to understand the cost of restoring one of my 2T drives.
It almost seems like the only use case for Glacier is for legal CYA reasons, where an astronomical cost of restoration still beats the legal costs of not having the archive to being with.
I suppose there is a use case for that. Sometimes it’s good for your piece of mind to know that your data is available and backed up.
I imagine there might be other use cases—for things where you expect a real-time response, as in when a user is waiting on data, it clearly isn’t valuable, but programs are more patient. If you have many asynchronous tasks, you can get a lot of work done very cheaply with this, especially if for the more common cases you hit a cache which can resolve more quickly, and only for the less common ones you hit glacier.