Currently Harvard uses Glacier via low-latency S3 tier, but has direct support for Glacier been in discussions in the community? It seems that handling long retrieval times of Glacier obviously requires extra functionality on the top of current S3 support.
I'm not seeing Glacier discussed much at https://groups.google.com/g/dataverse-community but you're welcome to start a thread about it!
@mjlassila we discussed Glacier and other chilled storage at the Dataverse Community Meeting back in June, and as you say a mechanism for handling response latency was the missing piece. many installations are struggling with handling big data well, and we would welcome more discussion.
On Harvard: it is my understanding that older/larger files are transitioned to Glacier by some lifecycle rule, but for truly big data they're now using Globus.
Last updated: Nov 01 2025 at 14:11 UTC