The Computer Backup Rule of Three
I'm ALWAYS pounding people to backup. I will continue. BACKUP YOUR STUFF. If you care about it, back it up.
Let's talk. Yes, I'm talking to you, non-technical friend. You're a writer, a blogger, not a techie. I get that. Can't be bothered, I get that. Very busy. You will be even busier when you lose access to your dropbox, or leave your laptop on a train.
Please. Read. Tell your friends.
These are NOT backups
Here are some things that are NOT backups. Feel free to tweet or Facebook them to
shame educate your family.
- Backing up your laptop to an SD Card in the same laptop is #notabackup
- Backing up to a hard drive that is 6 inches away from your computer is #notabackup
- Backing up your Gmail to another Gmail account is #notabackup
- Backing up your book by copying it to another folder is #notabackup
- The photos that are still in your camera memory are #notabackup
Do you have any other good examples?
The Backup Rule of Three
Here's the rule of three. It's a long time computer-person rule of thumb that you can apply to your life now. It's also called the Backup 3-2-1 rule.
- 3 copies of anything you care about - Two isn't enough if it's important.
- 2 different formats - Example: Dropbox+DVDs or Hard Drive+Memory Stick or CD+Crash Plan, or more
- 1 off-site backup - If the house burns down, how will you get your memories back?
Why so paranoid?
Simple. Because I care about my work, photos and data and I would be sad if I lost it.
Think about all the times you've heard about a friend who has lost everything. A decade of photos. Years of email. It hurts- just like exercise - because it's good for you.
Try restoring from a backup to practice. Backups always succeed. It's restores that fail.
What should I do?
I think at a minimum folks should do this.
- Have TWO physical backups (hard drive, memory card) with a copy of everything, at least weekly. You can automate this.
- Backup everything that has data that matters. That means phones, too.
- Have a cloud backup storage (CrashPlan, DropBox, SkyDrive, something)
- Don't trust the cloud. I backup my gmail, too.
- Rotate the physical backups between your house and somewhere else. I use the safety deposit box. You can use your Mama's house. Just label one "Backup A" and one "Backup B" and when you visit, swap them.
Here's some other blog posts on the topic of backup. Now, take action.
- A basic non-cloud-based personal backup strategy
- On Losing Data and a Family Backup Strategy
- Windows 8, Step 0 - Turn on continuous backups via File History
- Automatically Backup your Gmail account on a schedule with GMVault and Windows Task Scheduler
- Give Grandpa and Grandma the gift of an off-site backup of your photos
Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.
One I like for non-professionals (of the computer type)is CrashPlan. Like many other serivces it can backup to their cloud, but it can also backup to other computers that you designate. Users of the other computers can not (easily) access the contents of the backup so with a "circle of friends" you can each backup to the others machines.
Of course the "six inch" rule is not sufficient distance, nor is even the same houshold. Unfortunately there are disasters [think Sandy and Breezy Point NY] where entire neighborhoods get demolished....
Having to care about backups and backups of backups of terabytes of data is frightening. How can I backup 1 (or more) TB in the cloud or on DVDs?
PS: Ironically, last week I have finished setting up a FreeNAS 8.3.0 with a ZFS mirror RAID for my digital valuables. :P
Live and learn :)
Think of it like having to maintain a 1:1 ratio on BitTorrent, only it applies to everything you ever download from anywhere and you still have to do something about the data you had when you started.
I imagine these nice-sounding cloud backup solutions only work if you start when you're a kid and have no data.
Let me share a story.
One day in the not to distant past, some external auditors at our company find something that they need to take a "deeper" look at. And by "deeper", they mean to look at some financial records from about 7 years ago from a tertiary accounting system we had. Being so old and not used in any decision support systems, the data had thus been archived and sent to long-term storage -- as per requirements.
"No problem", say our accountants. "We'll just have IT restore that data from our archives so you can take a look at it."
So the request was sent to the storage and backup teams, and a large smile came across their faces. After all, this is what they train for -- retrieving data from the depths of our organization. They always kept multiple copies in multiple places; that was just common sense. They even kept spare copies of the format of the blank tapes in the vaults along with spare copies of the tape drives needed to restore those tapes. Why, they even regularly tested restoring the contents of those tapes onto disk just to make sure they hadn't degraded. In a number of hours later, we had a copy of all our data files dumped to a convenient location on our SAN.
Sounds good so far, right? Any clues as to the point of this story yet?
Well, here is when the trouble started. You see, in the last 7 years we've been on a track to modernize a lot of our IT infrastructure and software. This means for the most part, we have shiny new equipment, and the niftiest software for our users.
Upon notification to our DBA's and Application Support folks that the backup had restored they replied back: "Uhhh, whoops, this'll take longer than we thought -- it's not so simple to get at this data."
Two big things had happened in 7 years: We had upgraded our database from Oracle 9i on Solaris to Oracle 11g on Linux. We had retired that version of the accounting software. Some key staff members with domain-specific knowledge in the old software had moved on.
We had not only changed hardware platforms on the database side, but also endian-ess. When the application was retired, only online data was migrated into various new systems as part of a "one-time" project.
Obviously when this came to light, it was clear that it wouldn't just be a "plug-n-play" style restore that was presented as being possible.
In the end, once the backup was received, it required a lot of work and time by a lot of people to actually use the data once it was restored, even with the extra careful work done by the SB&R team originally.
The moral of the story, and one that home users and small-business might be interested in is this:
Ensuring the safety of your backups is half the battle. The other half is making sure you can read it once it's restored.
For home users, pictures and MP3s and such aren't much of an issue -- how often does the JPG format change, really? But will you be able to read that Microsoft Money backup 7 years from now? How about the files that neat program you use to jot down notes for later transcribing into legal briefs writes? Are those home videos encoded in some weird codec actually going to be watchable on a PC or TV in 7 years? How about 15? I recently looked at my backups and realized I would struggle to make use of some PCB layout and routing files from back in school, should I even want to.
Having well-protected and backed up data isn't worth anything if it all turns into the Voynich Manuscript in the end.
For the longest time, I had two HD that were mirrors of each other. NOT raided. About once a week, I would make sure the contents matched and went on my marry way. One was in my home server, the other was in a USB drive enclosure. My stance was that any event that would ruin both would take my whole house and that if they were printed (analog) pictures they would be lost too.
That was my stance, until this July. My house was hit by lightning. Lightning is a fickle SOB. It did not take the whole house. It made a hole in the roof and fried every electronic device on one circuit. When looking at the damage, I noticed the home server was not running anymore. Freaked out and found the USB HD that was its mirror, that didn't work anymore either...
By some magic or stroke of luck, both HD still function. The USB enclosure died and pretty much everything but the HD in the home server was dead.
I am still working out a backup system that has offsite and 2 copies in my house but as it stands right now everything is safe. Not organized yet but safe.
So what's you backup plan for all that data?
My new offsite is my car. When I remember to make my 3rd offsite backup, I just need to run out to the car to get it (which works as long as your car isn't tucked away inside your enclosed garage). It's a trade-off but at least the backups get done now.
BTW - in reference to a comment above, a problem with using SkyDrive synching to 3 machines is if you've ever done the Shift-Del on the wrong folder you can kiss all those files good-bye. You need at least one disconnected backup.
Jeff - Clever!
Joe - The Cloud is just one medium.
I'd love something that could automatically upload all my drives to Amazon Glacier (sadly Synology device doesn't support that yet).
I'd like to see a big button on the desktop or even on the keyboard (or perhaps automagically in the background) that backs up my work and encrypts it all in one go. A password/pass-phrase and button is as far as the typical user will go IMO.
There needs to be trust built into the cloud or there is no point. I bet users are willing to pay a little if the cloud they back up to can be trusted.
Andrew - See FileHistory in Win8, SyncBack for Windows, or TimeMachine on Mac.
Tomacsz - Google around, there's lots of work happening around Glacier.
Scott - I also pound on both my tech and non tech friends to do backups. What I've learned however is its better to get them to do something, anything, even if they don't achieve all 3 of your points. The 3 points will likely overwhelm the majority of non tech people. So even if they just did the lame back up to a drive that's 6" from the PC, its something - it's a start. Better than nothing.
That's why I wouldn't use Skydrive or Dropbox or Sugarsync or any of the sync services as backups. Backup services are distinctly different animals than sync services. I implement them both, but I don't view my sync service as a backup because if I did I'd be jumping through lots of manual hoops to try to get it to be such. Better to leave that to a real backup program and let sync do what it does best - sync.
Back in the 70's, I worked for a company that offered a mainframe software product that had its own proprietary database.
This story concerns one of our customers, the University of Toronto (Ontario, Canada).
Every night they ran an incremental backup of the database, copying everything that had changed in the last 24 hours out to tape.
At the end of the week (Sunday nights), they ran a full dump of all the database contents out to tape.
Then, to defrag the database, they deleted it from disk and restored from tape. This worked.
Until the day that, during the restore, the tape broke!
(The good news, such as it was, was that they were able to restore from the previous week's full dump, then apply the daily incremental backups. Since this was done on a Sunday night, they lost only the changes made on Sunday, a relatively light-usage day.)
The point here is that when the did the full dump, they had a backup of the data. Once they deleted the disk file they no longer had a backup!!!. What had been a backup copy on tape had implicitly been promoted to the Master. At this point, there was only one copy of the data.
Of course they changed their backup procedure. Backup the database. Copy the tape to another tape. Now delete the database and restore. (But arguably they should have done a Sunday night incremental dump as well.)
Several years ago, remembering the lesson of my previous anecdote, I had schooled a friend of mine to keep duplicate copies of her most important data onto two tapes.
At one point, I needed a copy of one of her files and (in the days before flash drives) asked to borrow tape #2 for a day as the easiest way of getting it onto my machine.
To my surprise, she balked at this. "But it's one of my backups. What if I need it?"
I said to her that, in all honesty, tape #1 was undoubtedly adequate, and that I never really expected her to need tape #2. The clincher was that if I actually did expect her to use it, I would have had her make yet another copy onto tape #3.
She saw my point and immediately handed me the tape.
As you point out backups always work, it is the restores that fail. If you don't test the restore don't plan on it working when you need it.
Or as my engineering friends say, one is none and two is one. Good point about having three copies.
Some may find CloudBerry (http://www.cloudberrylab.com) a good solution. Not only will it backup data to low cost cloud storage such as Amazon Web Service - Glacier storage which is 1 cent per gigabyte per month; it also encrypts and compresses the data before uploading so you do not need to worry about data breaches.
I back it up using ShadowProtect - hourly incrementals to our NAS. I can go back to any point in time and have a PC that I can boot (of course I also use source control, etc). A while ago we switched around storage and I didn't set up the ImageManager process in ShadowProtect so it wasn't collapsing those backups.
When the laptop screen died I thought "no worry" and figured I'd restore from the ShadowProtect backups to a VM and continue working. Having ShadowProtect collapse 400+ incremental images during a restore makes the restore VERY SLOW! Days slow...
I had tested backups by pulling out old files occasionally but never really tried the "disaster recovery". Lesson learnt without much harm for me fortunately. I just made a new ShadowProtect dump to a big USB3 HDD and am running the fruits of that dump in VirtualBox on my old laptop (nice that I don't have to restore it somewhere to run it).
If the new laptop is away for a little while then I'll migrate the image to one of our servers running Hyper-V (loving the new stuff in Server 2012!). I've lost enough time already but there's a weekend coming up I suppose.
Oddly enough I had just started last week to set up another dev environment since, even if you have code, if you can't compile a running release then the code's useless. My old laptop, which is also backed up, is able to compile all of the code, but after not touching that Windows instance for about 14 months I can see why I was happy to move to the new instance :P So much crud built up (experimenting with SharePoint 2010 and going "wow, I can run it on Windows 7" is a great way to waste space and resources even if the services are not auto-start).
Anyway, great post!
Then their disk crashed, and I helped them configure a new system. We came to restore from their backups, and found about three years' worth of accounts SOFTWARE but not a single byte relating to their data. Ouch.
Here is a summary of why
1 copy to external drive, 1 copy to a Thermaltake BlacX with 2 rotating drives (that go to my office), and 1 copy to my google drive.
I dig the BlacX since everybody and their mom has drives laying around to use for backups and also comes handy when attempting to help friends who don't back up their data and their system crashes.
Moral: If the data you need to save is organized and human readable, say payroll records from years ago, it might make senses to dump it out in that format. You wouldn't have to wasted a train-car-load of paper either. You could write it out as archive-version PDFs and save those several locations.
It might not be the best-of-all-solutions, but it'd be far less costly than maintaining or restoring ancient database software. And it'd be particularly helpful if there's only a handful of records you need to access for an audit.
Wrongamundo..an external drive is a legitimate backup. Won't protect you if your house burns down, but that is very rare. Will protect you for the most common scenarios (user error, hard drive failure)
"Backing up your book by copying it to another folder is #notabackup"
Wrong again, a 2nd physical drive in the same PC is a great option unless you hang out with people that carry powerful magnets.
"Don't trust the cloud?" #paranoid #misinformed
I am also a little bit "paranoid", although nothing ever really happened to me. Luckily, in all those 20 years since I am using computers I never had to use one of my backups. But sometime is always the first time, so I basically keep 3.5 backups. 3 "real backups": MacOS TimeMachine backups of my complete HD. 2 of them every day (1 at home, 1 in my office) and the third one ~1/year when I visit my family in Germany (I have one time machine backup across the ocean at my parents home).
And 1/2 backup on Dropbox. I use Dropbox for stuff that I am currently working on, when the project is finished, I move it to my HD Archive folder.
3. External drive
Also: sync between machines (using freefilesync + dropbox)
Plus send a spare drive to the parents.
DVDs and thumb drives too space-limited to bother with nowadays.
about ten years ago I backed up photos and important data on CDs and DVDs as I had no money for additional harddrives.
Then, someday after maybe two PC switches, I noticed there were photos missing in my archive on the PC. I started to search for them on the DVDs and CDs but three years of photos are still lost :(
I guess I missed to backup them (or lost the DVDs/CDs).
So, you should always be aware of what you backed up (like DerekTP mentioned above) - and where you backed it up.
So they proved that the tape backups that were being made regularly could be recovered from.
Then one day, someone actually needed something from the backup tapes. No problem. The tape was pulled from the bottom row of the rack, mounted, and... It was unreadable. No problem, they could go to the previous day's tape. It was unreadable. Previous day, unreadable.
What was going on? All these tapes were on the bottom row of the tape storage rack. They liked to keep the data center neat and clean. So every night someone came in with one of those big floor waxing machines. Also known as automatic tape erasing machines. The whirring electrical motor was not only polishing the floor, but sending out demagnetizing waves of electrical energy that zapped everything it could get near to.
That is an example of why having one backup in one spot may not be enough.
Remember when Megaupload was one of the biggest cloud storage providers? The Megaupload.com shutdown proves that an offsite backup service is not enough.
I rarely remove photos from my phone as an emergency.
I have a desktop with most of my photo/video/text work
Dropbox and Skydrive sync lots of stuff. Dropbox is the primary, but I'll copy over stuff periodically with scripts to Skydrive.
My laptop syncs this same stuff from dropbox/skydrive
Evernote contains lots of writings, but I export that periodically to dropbox.
Windows Home Server copies backups of machines, though since we're moving to Win8, I'm not sure how well that will work.
I still worry about an offsite solution for other stuff. I'm debating Crashplan/Carbonite/AWS-Glacier as solutions.
One other note, if you use Evernote/Dropbox/Skydrive. One of the issues is that if you delete something from your machine, it's gone everywhere. That's not a great backup plan. I've been slightly burned, and saved by killing a network connection here. I'd love to see some versioning or delay on delete propogation on these services.
This applies especially if you are using 3rd party tools to backup data into a proprietary format.
Always consider how to RESTORE data in the worst case..
and i haven't found any free service better http://goo.gl/OqQob
I use this windows backup software personally. The fact that it has unlimited storage was the selling point for me.
Great article, keep up the good work.
My laptop is my primary work tool - if it dies then I can't write any code, do any CAD, do any paid work... That is why I have an identical spare. The price of a second laptop soon pales into insignificance when I consider lost productivity waiting for my primary work tool to be revived. If one of the laptops dies I can move the SSD, or restore a backup OS image to a spare SSD (yes!), and resume work in no time at all.
Thank you Hg, Git and Dropbox for storing my data out of harm's way (mostly). The age of my OS images are relatively unimportant because I can get my latest data from its external home.
I use the second laptop for other things like client CAD work etc. I bought the spare as a factory ex-demo off eBay. The primary machine is the one with the 3-year on-site warranty. The hot-spare cost me less than a day's pay. Bonus - it doubles my compute power.
My offsite backup is a pair of 2TB External drives I rotate between home & a client's secure office. Speeds to and from the cloud in Australia are a lot more leisurely than in first-Internet-world countries so uploading to the cloud can be a (very) long-term investment.
Comments are closed.