a long time, I just had one copy of everything.

That's... not great. If there was a fire, all my data would be gone. Or what if Red

Shirt Jeff pulled one of my hard drives and ramset a nail through it?

Dramatic, I know, but in either case the data would be toast! So now I have a backup. And

actually I have more than one! Backups aren't just when you plug in a hard drive and copy

stuff to it. If you're doing it that way, you're doin it

wrong!

You might be saying, "Jeff, all my photos are in the cloud, and I do all my email, spreadsheets

and stuff online. My stuff's already backed up!"

WRONG!

First, the cloud is just someone else's computer, and second, you still need backups, and to

tell you why, let's start with the basics:

One, two, three.

Well, it's a tiny bit more advanced than that. It's the same thing, but backwards. Any data

you care about should be backed up following a 3-2-1 backup plan.

You should have 3 copies of all your important data. 2 of those copies should be on different

storage media. And at least 1 of those copies should be offsite, somewhere physically distant.

If you're relying on Google Drive or iCloud or something like that, that's just one copy.

You should still have two more copies, if you wanna be safe.

But what's important data?

Is it this scan of an old essay about computers from when I was in grade school? Is it this

old picture of my Dad teaching me how to ride a bike from the 80s?

Well, important data is subjective. It's different for everyone. But I'm sure you have some things,

like family photos or important documents you care about.

If you wanna make sure you don't lose those files, you need to cover all three parts of

the 3-2-1 backup plan.

And if you think you're somehow covered by using fancy RAID setups or ZFS snapshots on

a single storage array, you're not. Repeat it after me: RAID is NOT a backup! RAID is

helpful in many cases, especially if you have a huge amount of data to store, but it doesn't

matter if you have a NAS in RAID 10 or something like that—it still only counts as one copy,

and you need two more to be safe.

I'll show show you how I backup all my data, and hopefully you can get some inspiration

for ways to improve your own backups.

The most important thing I did before I bought my NAS I took an inventory, so I knew what

I was going to back up and how much storage I needed.

All my data can be lumped into six categories: photos, music, documents in my Dropbox, local

files outside my Dropbox, all my open source projects, and finally, my massive video library.

For each category, I follow the 3-2-1 backup plan.

My photo library, with over 70,000 RAW files, is more than a terabyte. I have thousands

of files from paid photo shoots, and thousands more of my growing family, so the library's

priceless to me.

I store one copy on my Mac locally, with all the original photos. Another copy's on my

NAS via Time Machine backup, and the third remote backup is to my iCloud Photo Library.

Some people hate it, but the convenience of something like Apple or Google's cloud library—assuming

you have an iPhone or Android—is life-changing.

In terms of trust, I know I'll get some sour comments, but I've been using iCloud since

it was iTools, and lived through the horrible abomination known as MobileMe. But you know

what? Because I also keep at least two other backups, I've never lost an image.

Moving on to my Music library, I have a lot of independent and children's music that isn't

on Apple Music or Spotify. I use Apple's iCloud Music Library and iTunes Match, and have all

the original files on my Mac and a Time Machine backup to be safe.

Important files like old schoolwork and business documents are in my Dropbox. That's all stored

locally, on my NAS via Time Machine, and on Dropbox. And for anyone about to comment about

how I should try out Nextcloud, don't worry, it's in the works. I haven't been too impressed

with Dropbox for the past few years.

Some other local files are only backed up on my NAS via Time Machine. So there's a small

hole here, but it's only temporary, since most of those files are things like OS images

that I could re-download.

My open source projects are a little more complicated. Almost all of them are on my

Mac, since that's my main workspace. That's backed up to my NAS, but I don't have every

project on my Mac.

Even if I did, my Mac might not always have the latest code, since I don't do all my work

on my main Mac.

So I have a separate backup on my NAS that uses Gickup. Cool aside: Gickup was created

by GitHub user cooperspencer after he saw me tweeting about code backups. It was really

cool to see this nice flexible Git backup tool created just from that Tweet!

So now I have one copy on my Mac, another copy guaranteed to be up to date on the NAS,

and a third copy, the source of truth, on GitHub.

Finally, I have all my video content. After this channel started blowing up last year,

backing up all my video projects got a lot more important!

I went from maybe having ten or twenty gigabytes of video a year to having five terabytes this

year.

Once I put my new NAS in service, I got more serious about backups.

I usually had two copies of most things, but they were spread out over a bunch of hard

drives, with no automation or documentation. But that's all changed. Now I have one copy

on my primary NAS, a synchronized copy on a backup NAS, and a third copy in the cloud.

And that's coordinated through this Raspberry Pi, my "backup" Pi.

I back up over 6 terabytes of data to AWS Glacier using `rclone`. `rclone`'s an open

source tool for synchronizing stuff to and from any cloud storage provider, like Backblaze,

Google, or AWS.

I store my backup in Glacier because it's cheap. Like, really, really cheap. I pay $4

a month to back up 6 terabytes. Before you rush to the comments, I know, restores with

Glacier can take a while to start and they cost more since they're offline, but you gotta

remember: the Glacier backup is for major emergencies only. Things have to be pretty

bad—like a nuclear attack, a burned down house, or a Red Shirt Jeff rampage—before

I'd need to use it!

But back to the Pi; it also runs Gickup, and the Pi logs everything so I can make sure

backups are running.

There are a thousand other ways to implement a good backup plan, and great software like

restic, arq, borg, and crashplane! I'm a video maker and developer, so my plan is probably

a lot different than yours. Maybe you could just get by with two external hard drives,

rotating them out in a remote location for the third copy.

But people tend to have a lotta data nowadays. Maybe that's you! Instead of relying on a

preinstalled backup app or a manual single-drive backups, you should take a data inventory,

write up a backup plan, and use robust backup tools to make sure you're covered top-to-bottom.

For my own plan, I created this open source repo, and it has my entire plan and includes

all the automation I use for my backups.

Even if this whole network rack got Thanos-snapped out of existence right now, I'd be okay since

everything's backed up and documented, and almost everything's automated.

But my plan's not perfect. An important counterpart is a disaster recovery plan. For a lot of

my data, I kind of wimp out and say "I'll cross that bridge if I ever need to."

But you know the worst possible time to test a disaster recovery plan? In the middle of

a disaster!

If my house burns down, I have to ask how long would it take before I could be productive

again, at least in terms of my data.

Would it take a day? A week? The only way to really know is to make a plan, and test

the plan. And I've only done a little of that.

To truly rest easy, you have to have great backups following a 3-2-1 plan, and test your

recovery process. I still have other holes in my plan, too.

Some devices like my network router and switches have custom configurations that aren't automatically

backed up.

It's hard to automate configuration or backups for most consumer networking gear, but I can

always do better. And I think that's the theme with backups: you can always do better. I

hope something in this video helped you fill a hole in your own plan.

If you're not backing up anything at all, at least go buy a hard drive and back up your

computer. Windows has Backup and Recovery, Mac OS has Time Machine, and Linux has—well,

a lotta different options—but most of them are easy to use. And if you're already using

a hard drive or NAS, maybe also back up to a cloud service, or rotate out two backup

drives.

We can all improve our backups—it's a never ending journey. The key is to find a place

you're comfortable. I'll conclude by saying there are two types

of people in this world: people who have lost important data before, and people who will

lose important data. In both cases, people with a good backup plan sleep easier at night.

Well, that is unless they have a nightmare about Red Shirt Jeff running around with a

nail gun!

Until next time, I'm Jeff Geerling.