One of the add-ins I had on my old Windows Home Server was a beta of Jungledisk that backed up my Photos folder to Amazon S3. To fill that role on my new WHS 2011 server, I’m trying Cloudberry Backup for Windows Home Server 2011.
Cloudberry offers a free trial, so I downloaded the Add-in package and installed. WHS warned me that the package was unverified, but I installed it anyways. After install, the app appeared in the navigation bar as “Cloudberry Online Backup” on the WHS Dashboard and in the list of installed Add-ins. The user interface of Cloudberry has five tabs that are easy to understand. Before I could do anything with it, I needed to give it the credentials for my S3 account in the Settings under “Set Storage Account”. It prompted me for a Bucket name, which I assumed it would create if necessary. It didn’t, and complained that the named Bucket didn’t exist, so I went into the AWS console and created it. After that, the account was added.
Because I’m not currently using this to back up critical system data, I went into Options and tuned the upload speed to 75KB/sec (about a quarter of my current upload bandwidth) and tuned down the process priority to Low. At this point, I was ready to Setup a Backup Plan, so I launched that wizard. I selected my S3 account as the target, and was given a choice between the Advanced (default) and Simple Backup Modes. Because these are just photos that I’m not keeping multiple versions of and don’t need encrypted for extra privacy, I selected Simple. I’ll probably use Advanced mode for documents or other sensitive data I might backup to S3 in the future.
I was pleased to see that the Backup Plan wizard defaulted to showing backup sources at the share level rather than at the physical drive level. There was an option to add the physical drives to the view. I quickly selected just my “Pictures” share. I used the Advanced Filter settings to specify just to back up files with .JPG, .PNG, .MP4, and .RAW filenames, mainly to avoid backing up system files like thumbs.db and desktop.ini that Windows throws in to picture folders. However, it looks like the default selection of “Do not backup system and hidden files” would help me there. I opted for no compression, to use S3’s Reduced Redundancy Storage (RRS), and to purge deleted files after 30 days. I chose a weekly schedule with a backup at 1AM Saturdays to reduce the load on the server. The wizard then created a service to run with Administrator privs. I liked the email notification options, but chose not to use them at this time. Then the backup plan was created and I set it to run immediately.
I was concerned as I watched the initial phase of the backup. Out of over 24GB worth of 10,000 picture files, the job status listed just 100 files and 1.8GB. However, the network utilization was on target and the performance impact to the server (an HP ProLiant MicroServer) was negligible. As I was watching the initial backup, I paged through some of the other tabs and found them straightforward and easy to understand. As the backup progressed, its number of target files and capacity increased, indicating that it wasn’t intending to cap out the backup job early.
I was very pleased to see that unlike Jungledisk, Cloudberry transferred the folder structure and files without putting them in containers. I was able to navigate the folders and see the pictures from the S3 web console. Very cool.
As I finish this entry, Cloudberry is plugging away at the initial backup and all indications are that it will work even better than my old solution! Recommended.