S3 Intelligent Tiering on existing Deep Glacier data

Thank you for the quick reply. I have a small follow-up question:

Option 2: This is a good approach for cost optimization. New objects would be placed in Intelligent-Tiering immediately, and the async tiers feature would allow objects to automatically move to the Deep Archive Access tier (equivalent to Glacier Deep Archive) after 180 days of no access, all within the Intelligent-Tiering storage class.

Do you mean here that using option 2 (async tiers in intelligent) totally removes the costs of $0.05 per 1000 items INTO deep glacier, AND removes the costs of $0.0025/Gb retrieved from deep glacier (bulk) ?

so to calculate costs benefits, I need to track the number of monitored items versus the cost of archiving and occasional retrieval using lifecycles and manual restores ?

That's right @Geert_VDW. Intelligent-Tiering (INT) adds the monitoring cost, but all the transitions and restore operations within the Intelligent-Tiering class are included in that fixed fee, except for separately requested expedited restore requests from the offline tiers.

However, note that functionally, the two options are not equivalent. When you rely on (INT) to auto-tier the objects to the archive tiers, that will only happen after the object hasn't been accessed even once in 180 consecutive days. If you use a lifecycle rule to transition the objects from INT to S3 Glacier Deep Archive, that will happen on the day you set it to happen, regardless of access patterns.

On the other hand, restore operations from the INT Deep Archive Access tier are more streamlined than with the regular S3 Glacier Deep Archive. When you execute a RestoreObject operation in INT, the object will be automatically lifted back up to the Frequent Access tier, where it starts the regular automatic tiering process from the beginning.

By comparison, when you restore an object from the regular S3 Glacier Deep Archive storage class, the archived object will remain in Deep Archive, but a temporary copy will be produced and kept available for the amount of time you configured when you requested the restore. If you want to move the object back to online storage, you have to make that copy yourself, make sure it doesn't get transitioned to Deep Archive unless you want it to, and delete the original object from Deep Archive (since a copy would now exist in INT or Standard), to avoid having to pay for both copies of the same object.