3-2-1 is a common practice for making sure data is safe but how do YOU backup your data?
Tell us the medium you use, the frequency of your backups and how often you access it. Most importantly include any horror stories of loosing data to the sands of time and why it happened.
Losing data horror story was storing things on external hard drives many years ago.
My backups are only ālocalā on a DAS in RAID 1 because 1) I donāt trust any cloud/server providers and canāt afford it anyway 2) have no friends with a server.
I have taken maybe 3 intentional backups in my life, and not in any kind of organized or deliberate system. They were prep for transitioning PCs or troubleshooting drives that might involve reformating. Every other ābackupā I might have is āaccidentalā in that it just resulted from copying stuff for the sake of reorganizing or moving between drives/machines, and then not cleaning up after.
I was always generally aware of the fact drives can fail, but had a young naivety of āitās so rare it wonāt happen to meā. I havenāt lost anything actually important, mostly because I donāt really have anything thatās truly āimportantā. But I did finally have an external HD fail, losing a bunch of accumulated media. Some of it at least a little sentimental, but a lot of it just accumulated random junk (old movie rips, pictures and video clips saved from the internet, etc). Right now I have my boot NVME showing signs of damage and Iām languishing in dealing with it, though I am at least formulating a plan to be better, if not good.
Oh also ālostā a drive due to encrypting it and forgetting the password due to setting it to auto-unlock on my own PC, then upgrading later. I still have the hardware for the old build, so in theory getting it back is a matter of rebuilding and hoping the boot drive (that I plugged into the new machine as a secondary drive) wasnāt messed up in some way that the encryption software doesnāt recognize it. I tried to be ācleverā with the password hint and have an actually strong password (also a practice I donāt really follow). Set the hit to a super obscure phrase/set of words that I thought āthis will definitely jog my memory when I read it and Iāll remember exactly the creative PW Iām using hereā and well, that didnāt work. I know generally whatās in the password, but canāt get the exact permutation right.
2 copies, and an ancient ābackupā āoffsiteā (read as in a detached garage 17 foot from the house)
data hoarding is expensive.
my first ābackupā was a USB HDD, that i dropped exactly after making my first backup. that caused a downward spiral of my sanity, and now i have a rack, ZFS in mirrors, and some what of a backup.
All data except /tmp and cache is in LVM RAID1 or BTRFS RAID1
Weekly: Automatic Cron snapshot of all my btrfs filesystems copied to an external HDD
Every 1-2 months: Sync those btrfs snapshots to an external HDD I keep at work, and rsync non-btrfs filesystems
Whatever drive has my financial data on it is always the drive to fail. Three times! Only data I lost was cache data, the rest always restored from RAID1 or btrfs snapshots easily
Iām currently reorganizing my setup and plan to upload a small <2GB set of extra-critical data to a friend sever in another state
Some of my stuff is mirrored or parity, but for actual backup I just use BackBlaze. Cheap and unlimited.
I lost a drive one time since I started backing up, it was a TB or two. Backblaze let me get a portable hard drive with my encrypted data on it, copy it off to my own drive and send the portable one back to them. No charge.
I currently store around 300TB or so on my Backblaze account, costs $9 a month for that much and I can still add however much more that I want.
Many moons ago I would just transfer my files from one machine to the next (Windows) and leave it at that. Then I had files corrupted, that I really wanted to remain intactā¦so naturally I waited 15 years before doing something about it!
I used to have 2 Windows machines, one was used for work/play and the other was just used to store old or big files. When I ran out of space, Iād buy another hard drive, but at least it was always backed up with BackBlaze or Iād periodically make DVD drive backups. It didnāt deal with the issue of silent corruption though
Fast forward to the last 3-4 years and with the help of L1 and Tom of Lawrence Systems, Iāve upped my data storage game.
I now have:
24/7 TrueNAS Server - contains hot files that change often (uses SyncThing & BackBlaze B2)
Cold Storage Server 1 - contains static files that change rarely and entertainment files (Films, etc.)
Cold Storage Server 2 - is a backup target for Server 1 and the 24/7 TrueNAS Server
Cold Storage Server 3 - is a backup target for Server 1
Cold Storage Server 4 - is a backup target for Server 24/7 TrueNAS Server
Cold Storage Server 5 - is a backup target for Server 24/7 TrueNAS Server
This is overkill, although Servers 4 & 5 are stored in a separate building on the same site, I do this because Iām not 100% competent. As Iām not in the industry and therefore donāt do this for a living, I like to have redundancy against myself!
All cold storage is turned on every 2-3 weeks and I run a scrub. By not leaving all the cold storage on 24/7, I save myself enough money to buy a decent graphics card every year (of course I donāt!), because energy prices.
I fell into having a data storage interest sort of at the worst time. SMR drives were suddenly a thing, and a bad thing at that. FreeNAS changed to TrueNAS and world energy prices suddenly ramped up (per KWh prices tripled).
I mainly use 4TB & 6TB Ironwolf or 14TB Toshiba CMR drives.
I was trained with daily backups kept for a week, weeklies kept for a month, monthlies for a year, and yearly backups kept indefinitely. The common 3-2-1 advice has always seemed tragically inadequate to me, and the lack of cheap backup media its only justification.
However, these days using btrfs I have monthly backups going back to 2017, and weeklies for nearly 3 years, on a few cheapish hard drives. This is because using btrfs incremental send/receive, the snapshot history is preserved on the backup volume. The weekly backups go to a rotating set of drives, and ideally these get rotated off site (er, thatās from home to a family memberās place). When a backup volume fills up, I try to buy a new one, but sometimes Iāve had to prune old snapshots.
As well, thereās hourly automatic snapshots and daily-ish send/receive to an internal hard drive.
This has needed a bit of curation of the data, mostly putting caches, downloads, and scratch data into separate subvolumes. The browser caches were the worst, gigabytes of rapidly churning data.
I have the Veeam community server running on my central media PC with agents deployed to the machines I care about. That is then offloaded Backblaze for my 3-2-1.
My retention is pretty narrow since we donāt have a lot of data that changes very often. Everything we have on our PCs today is what we need tomorrow. I just havenāt run into a scenario where I need to restore past 60 days ago.
Honestly every says this but almost no one does it. I have never actually seen anyone use 2 medias. This seems to me like advice from when tape made more sense. Now days I mostly see people just using two servers, or a server that is cloud backed up, and I wouldnāt consider the cloud a separate media. Since it just ends up on HDDs anyways.
Only time ive ever lost data at work was doing a large bacup
Power surge wiped out the pc, hdd and backup drive and the docking station.
They did not have a surge supressor/ ups.
( cheap boss)
There have been two big events in my life that have led me to where I am regarding computing, security, and data integrity. The first was the time my Dad, a practicing Psychiatrist, got a virus on his windows PC that started emailing out random snippets of random files, some of which were patient notes and files. If youāre not aware of the consequences of HIPAA violations, go look it up. That was when he, and I, started taking data security seriously. The second was when I, an ignorant Apple user thought that having stuff on a Time Machine backup meant it was safe forever. I was wrong and I lost a lot of important memories when that drive died. Yes, I was able to recover most, but at a cost.
Now, I am an advocate for ZFS, even on desktop systems, and making an OS that runs on ZFS and is easy enough for my now elderly parents to use.
My backups are a brand spanking new HL15 that runs TrueNAS Scale and that backs up to Backblaze B2. My Brother is still soldiering on with a TrueNAS Mini XL and we share space on each others NAS for yet another backup location.
I hope that is enough, barring nuclear war or a giant asteroid hit.
After a forest fire almost took out my house, I got even more serious about backups, especially off-site.
I use btrfs, so I use btrbk to create daily snapshots of root and home and send/receive those to a second internal drive as backups. I also run restic to back up my most important files to Backblaze B2. Then finally, I occasionally use CloneZilla to create a full image of my drives and put them on a USB drive I keep at work. At some point I suspect Iāll add a NAS to the picture.
About 5 years after I started my photography business (many thousands of photographs in my catalogue) the data drive failed in my Dell workstation. No problem, installed a new data drive in the workstation and tried a restore from the daily backup to my Drobo. Ugh, the Drobo failed, literally died! Get the offsite monthly backup hard drives. First drive failed whilst restoring. Iām now down to one copy of 5 years worth of work. Luckily that drive held for a restore.
Iām now using a Synology NAS (RAID 1) backed up to their cloud backup service, and have two monthly backup drives stored in the home office at the far end of our garden.
The Synology has been solid, and Iāve done a few test restores from it, although I donāt do regular restore testing
I keep an eye on Synology to ensure that the quality of their products and services remains high. If it drops I would change, but in ten years and a few different Synology NAS they have been v reliable.
It probably also worth saying, I no longer have that business, and the above backup is for personal data and the photography assets from that business. So the backup restore requirement is quite light.
I store things on mirrors on the main system, for some redundancy.
Periodically (usually 3 times a year), I compress everything important and save it to a new folder in an external HDD. If the remaining space is too small, I buy a new one and keep the old ones.
Thinking about adding a āNASā to this equation, which in itself wonāt be a backup, but will be another location the data exists on. (Though ZFS+no ECC fears are real/ price for alder lake ECC systems is high.)