I propose to add some functionality to git-annex, to automatically "throttle" the data retrieval from amazon glacier to stay within the daily free retrieval allowance. If someone would need to get his/her files faster, there should be an option to disable this throttling (or even better, specify the retrieval rate).

As far as I understand glacier, this could be implemented using range retrievals. In short range retrievals enable you, to only retrieve a part of an archive in glacier. This can be used to only retrieve / request so much data, that you stay within the free retrieval allowance. ( please see Q: Why would I retrieve only a range of an archive? ).

This would be somewhat similar to the smart retrieval feature in cloudberry .