If you’re looking for cheap offsite backup storage and are ok waiting up to 24 hours for the ability to restore these backups, Amazon Glacier seems to be the main contender out there. At .01 per GB per month and no cost for storing or deleting backups (they do nick you on restores), it’s hard to beat.  The biggest problem is the lack of backup software that will write to Glacier, especially if you want to automate the process, and again worse if you’re backing up from a linux host.  Recently I was tasked with just that, getting backups set up for a linux server that we host on Amazon and storing as many of them as we could for as long as we could, for as cheap as we could, and our return to service SLA is around 48 hours.  Glacier is the obvious choice, beating out all other cloud storage vendors by almost 10x (Amazon S3 and Windows Azure came in at the same price for the first TB, .095 per GB per month), but I was having a hard time finding software to use for the backups.  Eventually I came across amazon-glacier-cmd-interface on Github (I know, using development software in a production environment is a big no-no, but I figure as long as I keep an eye on it we should be fine, plus we keep 2 weeks worth of backups local anyway), it provides a relatively easy command line interface to backup to and restore from Amazon Glacier, though the documentation leaves some to be desired.  Thus, my post. After spending quite a bit of time (I’m not terribly linux savvy) on a test server getting everything working, rolling out to a QA server for verification and then rolling into production, I’ve worked out almost all of the kinks.  Below is my entire process for installing amazon-glacier-cmd-interface along with all of the prerequisites and getting it configured and working.

Step 1: Install Git

Git is the program used to clone projects from Github to your linux server and though it’s not required (you can download the program/script file separately and unpack on the server) it makes installation much easier.

Step 2: Clone and install Boto

Boto is a required set of scripts that amazon-glacier-cmd-interface uses as a backend to talk to Amazon Glacier.

Step 3: Clone amazon-glacier-cmd-interface, install prerequisites (Glacier Core Calls and Distribute tools) and install amazon-glacier-cmd-interface.

Then:

OR (depending on your version and distro of linux.  Distribute utility replaced setup tools.)

Then:

At this point, amazon-glacier-cmd-interface is installed, now we must create the config file: /etc/glacier-cmd.conf.  Here is a sample:

At this point, you’re ready to start backing up to Glacier.  The following command is a sample backup command.

Also, here is a sample backup script.  This script will tar your backup folder, calculate the amazon hash, upload the tar file to Glacier, and delete the tar file (but leave your backup folder alone).

After that, all you need to do is add the following line to your /etc/crontab.

And your offsite to Glacier will run every morning at 1.  I’ve had pretty good success, it’s been installed for 2 months backing up our environment and I think it’s errored out on 3 backups (which I was able to go back and manually upload the next day).

It’s not the most elegant solution, and I would never consider it to be an enterprise-level solution, but in the absence of an enterprise-grade solution, it will do. A few things that I didn’t cover:

1. Multipart backups – Amazon Glacier requires multipart uploading for anything larger than 4GB, but recommends it for anything larger than 100MB.  Our backups are about 2.5GB, so I haven’t had to deal with it yet.  I know amazon-glacier-cmd-interface supports it, but I haven’t looked into it.

2. Restoring backups – I’ve attempted several times to restore using amazon-glacier-cmd-interface but have had no luck, it basically blows up on me.  What I’ve resorted to as a restore process is using Cloudberry Cloud Explorer (there’s a free version) to pull the backup file down, then FTP it to my linux server and unpack it.  Again, not elegant, but it works.

NOTE: I had some real issues getting Cloudberry Cloud Explorer free version to connect to Glacier.  It would connect to S3 just fine, but would error out while connecting to Glacier.  After working with their support, we were able to figure out that if you install a trial of the Pro version, then uninstall the trial, and then install the free version, it will work just fine.  Apparently there’s a registry key somewhere that it needs to connect to Glacier that the free trial is missing.  Installing the Pro version gets it there and then leaves it there during the uninstall (something I usually hate about programs, but actually worked out in this case).

3. Catalog – amazon-glacier-cmd-interface does have the ability to keep a catalog of your backups to simplify the restore process.  It utilizes an Amazon SimpleDB database, where it stores information about your backup sets.  I looked into this a bit, but it seemed to overcomplicate things for what I was looking for (and also for my experience level to be honest).  I did not work with it, but I thought it was worth mentioning that it does exist.

 

Download and Docs for Botohttps://github.com/boto/boto

Download and Docs for amazon-glacier-cmd-interfacehttps://github.com/uskudnik/amazon-glacier-cmd-interface

Download and Site for Cloudberry Cloud Explorerhttp://www.cloudberrylab.com

 

 

 

 

 

0.00 avg. rating (0% score) - 0 votes