Every several months I consider re-jigging my method of backing up my NAS at home. I have several TB of files, mostly images. What I do now is basically only backup my exported, “finished product” images as jpegs, and periodically back those up to a pair of spinning hard drives, with one of them being in the safe at home, and the otehr in the safe deposit box at the bank, with the drives being rotated periodically.
I’d like to back up to a more robust media, something actually designed for offline storage. As I understand it, that basically leaves me the choice between some form of optical (blu-ray and m-disk, basically), and tape backups. Tape backup systems are quite expensive upfront (well at least the ones that are near-current-generation), but it has the added benefit of systems being available to handle multiple tapes, since the amount of data I need to back up won’t fit on a single disk…
Could it ever be good bang for the buck?
Lets say you have 2.5 TB and pay $144/TB/year to store it (rsync.net pricing), totaling $360/yr for an always-live backup. On the other hand, something like amazon glacier deep archive can cost as lows $12/TB/year, totaling $30/year for your data, to store backups that you access only in a disaster recovery scenario. There are additional costs for uploading and downloading data with glacier and comparable products.
Say you can fit all of your data on 25 pack of 100GB M-Disc drives, amazon says that costs $245. Not including the time and effort it takes to back up and organize twenty five optical discs, you can make at most one backup per year before it’s cheap to go with an always-live option and never cheaper than a deep-glacier-type backup.
Let’s say you buy an LTO-5 drive. Newegg says you can buy a low end unit for ~$800 and that the drives cost about $35 for a 3TB tape which you can fit a full backup on. Supposing you make two backups per year you break even on the always live backup in about three years and never cheaper than a deep-glacier-type backup.
The math shakes out differently for each service, but it’s hard to imagine a scenario where its a good bang for the buck on the order of a few TB.
I’ve waffled on this topic a lot over the years. The cool factor of LTO is appealing from a selfhosted perspective, but the reality is that, unless you’re willing to spend a LOT, or are willing to go way back on older-generation used hardware, it doesn’t seem to pan out. Plus I hear the used enterprise gear is pretty loud, so that’s an additional negative, unless you go for a desktop solution, which are really hard to find cheap.
(that being said, look how pretty… EH970A compatible LTO6 external tape drive LTFS (inside view limited edtion) | eBay)
It kinda seems to me like purchasing a few Manufacturer Recertified 20+ TB hard drives and doing manual / semi-manual drive swaps either with a desktop NAS or desktop dock, then cycling the drives physically out to a safe space is the way to go for the “must be safe” kinds of things. Having multiple allows you to have peace of mind with the case that one goes bad in storage, you’d still have other(s).
Beyond that, my photos are synced to backblaze, since I really don’t have all that much, so it just is easier to sync that from my NAS to the cloud and pay for it.
I recently moved everything to Immich for photos, and while I’m still trialing the software, it seems fairly good, so I’m beginning to look at backing that up first to my NAS, and then potentially using a cloud account (probably backblaze again, honestly) as the third location in the 3-2-1 strategy, but I’ll have to investigate it more closely. Since Immich is a VM on proxmox, I already back up the whole VM to my NAS with everything else from Proxmox, but Immich also has the option to just backup the files and database without the VM, and I have to decide how I want to handle that.