My 3-2-1 Home Backup Solution: Drobo 5N + Arq Backup + Amazon Glacier
I've been backing up my home data to a Drobo 5N NAS for a few years now, but never felt fully comfortable about the idea that if my home were to burn down, that data would be gone.
I messed around with CrashPlan before, thinking I would take advantage of the unlimited space but I think they stopped offering that. Also I wasn't able to get the headless agent on the machine and the steps were just a little too inelegant/homegrown/unsupported. Still, at the time it was the only option and I figured I'd get to it eventually when I had time.
On a whim, I started researching alternatives this holiday and I'm glad I did. Within a day, I had a better, much more elegant solution that came in the form of Arq Backup. Combined with the low cost of Amazon Glacier whose pricing model and design fit my needs completely (ie., I only need to retrieve my data if my house explodes and I'm still alive), this really provided me with the exact solution I'm looking for. All for a one-time upload cost of about $50 and a per-month cost of $2.40 for 600GB of data. (Check out this calculator I made to figure out what your cost might be)
Here's how my backups work.
First & second copies: on my computer, iPhone, what-have-you
I have things on many devices around my home and in my mobile devices. As soon as I enter my home, I can connect to the network and store these files on the Drobo 5N. Here are a few ways I do that:
- Sync files from my Mac computer to the NAS on a regular basis. I use a program called GoodSync which has reliably been copying specified folders to my Drobo 5N at regular time intervals. Things like tax folders, creative and client work, etc. just get synced without my having to think about it.
- Transfer all photos from my iPhone to the NAS. As soon as I walk in the door, my phone also does an automatic transfer of any new photos or videos I've taken to the NAS thanks to a utility called PhotoSync. It keeps track of photos and videos that I've already uploaded and only puts the ones that are new. Unlike GoodSync, which communicates with the Drobo via a mapped mounted volume, I had to set up an FTP server on my Drobo for TransferSync to be able to connect. TransferSync is able to then use the FTP protocol to upload the files.
- Access and save anything on any iOS device. Using FileBrowser, I can browse any file or stream any photo or video on the NAS using any iOS device that I have, as well as save things to the Drobo. This helps me feel okay about deleting all photos and videos on my iPhone, especially when it starts to get full. I won't be able to look through photos while I'm away from the house is the only drawback, but I usually go through old photos while I'm at home anyway. And I keep a subset of favorites on my iPhone.
Third offsite backup: Amazon Glacier via Arq Backup
So copies of my files are on my various devices for immediate access, and on the Drobo 5N as a backup. What about if the house burns down?
I currently have Arq Backup encrypting and copying all the files on my Drobo 5N to an S3 bucket in the Amazon cloud, and setting the lifecycle policy of that S3 bucket to make it a Glacier bucket. For the uninitiated (ie., me just a few weeks ago), what this means is:
- Arq Backup is a client that sits on the computer that has access to the NAS. It bridges the networked drive to any cloud service— in my case, it's a bridge that manages the backup between my Drobo 5N and my Amazon Glacier storage. It supports all kinds of cloud services though, like Google Drive, B2 BackBlaze, etc. (but not iCloud).
- Amazon Glacier is a specific storage offering by Amazon. It's a cloud storage similar to S3, but the pricing schedule is much different because unlike S3, Glacier is designed for archiving data that will be very infrequently accessed, which is perfect for me. The storage prices per GB per month are greatly reduced compared with S3, but the cost to retrieve and download are much higher both in terms of time and money. This is fine for me because I don't ever plan on retrieving this data from Amazon Glacier unless, again, the worst happens to my primary and secondary storage locations (devices and Drobo 5N). I had a lot of issues calculating what my storage costs would be for Amazon Glacier, even using this unofficial Amazon Glacier cost calculator. I really wanted to know given that there would be a lot of transfer activity with initially uploading the data and also executing what is called a "lifecycle transaction" where I am converting S3 buckets to Glacier buckets. So I made my own Amazon Glacier cost calculator in case anyone else wants to see what their costs would be if they have my exact use-case which I imagine would be pretty common.
For 600GB of data, this is probably going to take 60-100 days because we have slow-as-molasses Internet, it takes up all the bandwidth when it's doing its thing, and so I can only do it from about 11pm - 7am every day. Which is fine. This is just the initial time investment before I can get to the point where everything is just incremental backups of infrequent files and photos.
I also tried restoring one file from Amazon Glacier, and it worked like a charm. So this helped ease my mind.
This remains to be seen in fact, but theoretically this setup is also future-proof. I had a concern about how Arq Backup is encrypting my data. If Arq Backup goes kaput one day, and I somehow lose the computer that I have it on and it's difficult to get another copy of Arq, how will I retrieve and decrypt the nonsensical data that is on Amazon Glacier?
The creator of Arq put my mind at ease on this front. He said the encryption algorithm used is open source, and that there exists an Arq Restore command-line utility which is also open source that will be able to retrieve files. So I really only need a Linux distribution that can run this command-line tool and I will be able to access what I need. Given everything Amazon doesn't blow up as well.
So, I'm pretty happy. This is better than what I had before, and got something off my to-do list!