Amazon Glacier + Lightroom ???

[Alan Goldhammer]

@cybis - this was my feeling as well from reading the documentation.  The lack of flexibility for doing incremental backups is a the big one for me.  As I noted in an earlier post, I use Mozy Home for cloud back up and although I certainly have to pay much more than I would for Glacier it's user interface and ability to do the incremental back ups is just perfect.  Software is automatically configured to back up twice a day which may be a little redundant.  One can do direct file management as well and they now have Android and Apple OS aps that allow you to access files as well though I don't think I would ever need that but who knows.  I also have immediate access to cloud files for downloading if necessary.  It's important to remember that uploads are really the time consuming thing and if you are forced to do big uploads on a regular basis as it looks like with Glacier that's going to take time.

[cybis]

I hear you, Alan. Though I think it should be possible to have the best of both worlds by compromising, at least until a better front end for Glacier comes along.

By splitting the data in two, the cost saving of Glacier and the flexibility of traditional cloud storage can be achieved.

Glacier/S3	Savings
90%	81%
75%	68%
50%	45%

Moving forward, I don’t see why a better front end wouldn’t be possible. All that is needed is to keep track locally or in an ‘earth’ cloud of what was uploaded to ‘Neptune’ (Glacier). So this smarter system would be able to perform incremental backups. It would know if a file already exists, is new, or updated and take the appropriate actions from there.

[Alan Goldhammer]

Moving forward, I don’t see why a better front end wouldn’t be possible. All that is needed is to keep track locally or in an ‘earth’ cloud of what was uploaded to ‘Neptune’ (Glacier). So this smarter system would be able to perform incremental backups. It would know if a file already exists, is new, or updated and take the appropriate actions from there.

I think conceptually that this is pretty easy to do.  Looking at my own situation, I have 68 GB of images and my LR catalog is 77MB.  Uploading the catalog regularly isn't going to be a big deal because of it's moderate size so I think that we don't have to worry so much there.  With the image file, the front end needs a database module to track when files are uploaded to Glacier and where they are located.  I do this the first time and name this "Glacier 1"  Subsequent uploads don't need to be concerned with any unmodified files that are in "Glacier 1" and only have to upload them to "Glacier 2" and so on.  Catalog file updates would be linked in the database just as they are by LR today in that backup system.  It still pretty kludgy but as you note, cheaper than current alternatives, certainly in my case with Mozy Home.  Now the big worry here (and maybe it's not really a worry) is the worst case scenario, a full computer crash with a loss of all data including your Glacier database.  How do you go about making a sensible restore?  I suppose you could put the database up in Glacier as well but it is certainly less efficient than Mozy at this point which totally mirrors everything on my hard drive in terms of file structure.

[cybis]

You lost me a little with Glacier-1 vs 2. Why upload new files to a different vault?

I'm thinking whatever future LR module or standalone program will keep track of all files in a database as you mentioned. The database itself should be stored in a fast access cloud like Mozy or S3 but not Glacier.

There would be only one vault, say Glacier-1.

When you click ‘backup’, the algorithm would then do the following:

• Determine if there are any new or updated files
• Upload all new and updated files to Glacier-1 (including the original path)
• Optionally delete previous versions of updated files (you would probably want to opt out of that option though)

   
When you click restore, it would compare the local storage to the database and download the needed files.

[Alan Goldhammer]

If I am staying with Mozy for my catalog storage, I don't need Glacier.  My understanding is that only packets can be uploaded to Glacier and these are discrete entities.  Maybe my language wasn't terribly clear in the previous response.  My point is that you will have a collection of files containing multiple images and you have then have to download an entire collection if you need to access a single file that might have been lost.  With Mozy I can just download the single file.

[cybis]

You can upload and retrieve individual files in glacier.