Using Glacier with Cloud Gates

First thing you need to know about Amazon Glacier is that it is very different from other cloud storage -- it was created for archival storage and using it for any other purpose will not be a pleasant experience. Glacier is mostly concerned with uploading new files. Retrieving data from the storage and even inspecting what data you've previously uploaded is a much trickier process that requires waiting from 4 to 30+ hours. It's simply impossible to download data from Glacier without at least 4 hours prior notice.

Right now, for best results you should be uploading a few very big files instead of multiple smaller ones. If you want to use it for photo storage, start with creating a few archives with all the photos you want to store grouped by year for example. This will work much better than uploading thousands of photos as-is. See future directions section below for information on using Glacier with full folder structure.

Files you upload will be transferred to Glacier immediately, but you will not immediatly see archives you've just uploaded in the FTP folder. This is because the uploads take some time to process on the Amazon side. If your upload to FTP succeeds it's a guarantee that it's on Glacier as well and it will appear in file listings in about a day.

Upload to /incoming

This is where you upload your files. You cannot create subfolders, so if you need to upload a directory structure, zip it up and upload the archive here. If there are any unfinished uploads, you will see them here and will be able to resume your upload from where it was interrupted.

Once the upload is complete the file will be "consumed" and disappear from this folder, only to reappear in about a day in the /archives.

You can upload files up to 640Gb big (but also subject to your subsciption plan limitations).

List & delete from /archives

Once processed, your uploads will appear in this folder. You can list them, you can delete them, but you cannot download them from this folder.

The listing in this folder is lagging from the real state of the Glacier storage by about 24 hours. This means that files you've deleted will still appear here until the next Glacier inventory that only happens daily.

If you haven't looked at your Glacier storage for a while and this folder is empty, that means that there are no cached inventory results. When this happens we request an inventory behind the scenes and you should come back in about 4 hours and will see this folder populated again.

If you have uploaded files with the same filenames multiple times you'll see duplicate entries in this folder. Deleting one of them will delete all of the files that share that name.

Downloading files from Glacier

As we've mentioned before downloads are only possible with prior notice. Here's how to do it:

  1. In the root of the FTP server there's a file named downloads.txt
  2. Open it with a text editor and you will see a list of all the archives in your Glacier vault with some additional info
  3. Find the file you want to request for download in this list
  4. One of the lines pertaining to this file will say "Requested for download: no"
  5. Edit that line to say "Requested for download: yes" instead
  6. Save the file and upload it back to FTP if necessary

Once you do that it's in Glacier's hands. It will take about 4 hours to actually retrieve the file from it's "cold" storage and make it available for download. Once that happens you will find it in the /archives-ready folder.

Downloading archives from /archives-ready supports resume, so even if your connection is interrupted, you should be able to download even the biggest archives.

Future directions

We are looking to improve on top of what the Glacier storage offers and are considering a different type of gate that would have the following properties:

If you find this kind of Glacier gate more interesting, please express this by sending us a quick email. That will help us prioritize this work accordingly.