I've recently discovered that the older Glacier service is phasing out for new users. With all the updates and new Glacier classes in S3, it seems to be a sensible time to consider migrating my data. I'm particularly interested in moving to S3 Glacier Deep Archive because it's cheaper than the traditional Glacier, but I have several concerns:
1. **Cost**: Retrieving archives from Legacy Glacier with the "Bulk" tier is currently free for me, but there always seems to be a cost for S3 Glacier Deep Archive, even for bulk retrievals.
2. **Retrieval Time**: Legacy Glacier Vaults retrieve data within 8 hours, but I'm currently testing S3 Glacier Deep Archive, which claims retrievals can take "up to 48 hours". I'm already waiting for 34 hours on my first retrieval, and I'm worried about the long wait for a service that's supposed to be more expensive.
3. **Max Object Capacity**: Some of my archives are massive, including entire disk images. S3 has a 5TB limit per object, whereas I believe Legacy Glacier allowed larger files without a strict limit on size, though multipart upload constraints existed. Can S3 handle my large archives?
4. **Older SDKs**: I'm using older versions of the AWS SDK for automation, and I'm unsure if I need to upgrade, especially regarding support for Glacier Deep Archive.
Additionally, will I still be able to create new archives on Glacier beyond the December 15, 2025 deadline?
3 Answers
If you're aiming to keep things similar to your current setup, I’d recommend sticking with Glacier Flexible Retrieval. It's more suited for cases where you occasionally need to access data without long waits. If your usage is infrequent and you're okay with longer retrieval times, then S3 Glacier Deep Archive could be a good option since it's cheaper.
To answer your concern about the SDK: if it includes the GLACIER class, it might not have the DEEP_ARCHIVE support if it’s older version. It's probably best to update. And yes, for large files, S3 can manage files up to 5TB. If you're worried about the immediate need, you could start uploading data to S3 using the STANDARD class before transitioning to DEEP_ARCHIVE later using AWS CLI.
Just to clarify: S3 has a 5TB max limit for single objects, which it’s had for quite some time. Regarding your SDKs, S3's Glacier support has been around for a while, so if your SDK is older, a quick upgrade might be wise.
Thanks for the heads up! I was unsure about the limits and it's good to know I should be fine with files under 5TB.
Exactly! You should be good as long as your files are under that limit.
Got it, I'll consider that route for my urgent uploads! Thanks!