Backup : comparison and ideas

(Note : I recently got Internet access back, but only at work. That’s because my girlfriend is currently spending a week of holiday with me, and it would be rude of me to abandon her in order to write lengthy blog posts while I’m at home. So news on implementation and SVN commits, for which I need my own computer, still have to be delayed a bit…)

Backup is one of the most boring regular maintenance tasks of a computer. It is necessary, however, if you care the tiniest bit about the data you put on said computer. In this piece, I wanted to make a comparison between various ways a desktop computer user may make copies of his data, with a focus on what differs from the point of view of said user rather than inside the hidden internals of the hardware and software being used. I’ll also add to this comparison an idea I had recently on the subject, that I call “cross-backup” and which is to the best of my knowledge new in the realm of desktop computers, though larger server systems could be using it already.

Characteristics of a backup method

From the point of view of UX, the ideal backup method would have the following characteristics :

  • Low financial cost : If it’s crazily expensive, no individual user will bother buying it.
  • Easy setup : While in some scenarios (e.g. entreprise networks), people may tolerate having a third-party computer enthusiast taking care of their machines, for personal and home computers it’s best if users can take care of their data on their own.
  • Everyday comfort : The ideal backup method does not ask much from the user, so that there’s a lower risk of backups being forgotten, or postponed because they’re too boring.
  • Easy data restoration : We do backups because we fear the day where our computer will stop working and we won’t be able to extract the data from it. When that day happens, being able to get back to work quickly is a big productivity plus, that would be greatly appreciated.
  • Control on data : The data on our computer is partly personal, so we’d prefer not to hand it to unknown outsiders if we could avoid to. Who knows what they could do with that…
  • Robustness : Even with backups, there is still a risk of data getting lost, given that some conditions are met. This criteria evaluates how resistant this backup method is to failure.

Backup methods being considered

I’ve tried to cover a pretty large and representative set of data duplication methods, which I will now describe here :

  • External hard drive : The traditional backup method of home computers. Plug a USB/eSATA drive inand backup software will save your data in the background while you continue to do your stuff.
  • RAID 1 : Double the number of hard drives in your computer and have it automatically duplicate any data that gets written on the extra drive(s) you added. Other forms of RAID exist if you are ready to add up even more hard drives, which provide different options on the speed/reliability compromise.
  • Extra partitions : Partition your hard drive so that some regions of it are reserved for data duplication, and ask backup software to save its stuff there.
  • Network Attached Storage : Have a little server on your home network, providing data storage facilities, and use those to store a backup of your data. Apple’s Time Capsule is an example of NAS that’s dedicated to this.
  • Public “cloud” services : A growing number of companies provide their customers with some space on a web server, that they can noticeably use to backup their home PCs, for a variable price.
  • Robust server and thin clients : Have a big, heavy server, with hard drive content that’s duplicated multiple times and impressive computing power, control all computing power within the house. All other computers are in fact thin clients, which provide RDP access to an OS running on the big server, either directly or over a VM if several different OSs are required.
  • Cross-backup : Computers within the home network backup their data on each other’s hard drives silently when they are turned on at the same time.

Individual analysis and comparison

External hard drive
  • Low financial cost : An external hard drive of good capacity (ideally that of the HDD to be backed up) is required. It shouldn’t cost more than 100€. Average.
  • Easy setup : Backup software bundled inside of modern OSs makes it a no-brainer. Good.
  • Everyday comfort : No-brainer too, however the need to find the disk and plug it in every time makes things more annoying. Average.
  • Easy data restoration : A full OS installation and length copy procedure is required. Bad.
  • Control on data : You own the external drive and have as much control on it as on any other belonging. Good.
  • Robustness : This method encourages you to keep the drive close to the computer, so major incidents (fires, etc…) are likely to destroy all copies of the data. Still, that’s pretty good protection already. Good.
  • Low financial cost : Same as above, but with an internal hard drive that’s even less expensive, though the difference is small for laptops. Average.
  • Easy setup : Setting up a RAID is a thing for geeks, no one will dispute that. Bad.
  • Everyday comfort : Backup is done on the fly during every disk write. Good.
  • Easy data restoration : Finding and replacing the faulty drive is all it takes for the PC to boot. It takes a bit of computer hardware experience which not everyone has, though. Average.
  • Control on data : Same as above. Good.
  • Robustness : The hard drive is still inside of the PC to be backed up and exactly mirrors every disk write, so the only things which RAID protects against is natural hard drive EOL. OS failure, falls and other physical damages to the computer are still likely to damage the duplicate disk as well as the original one. Average.
Extra partitions
  • Low financial cost : Only costs some HDD space, which nowadays is essentially free. Good.
  • Easy setup : Partitioning is a bit more widespread knowledge than RAID setup, and could be done at most local computer stores or by geek friends, but it’s still not for everyone. Average.
  • Everyday comfort : Common backup software can be set up to write there, and you have nothing to plug in as a bonus. Good.
  • Easy data restoration : Requires an OS reinstall. Bad.
  • Control on data : Same as above. Good.
  • Robustness : Only protects against OS failure, and lightly so. Bad.
Network Attached Storage
  • Low financial cost : At a few hundreds of euros, NASes tend to be priced more like a computer than like an external hard drive. Bad.
  • Easy setup : Good backup software can make it a no-brainer (cf Apple’s Time Capsule for an example of easy setup). Good.
  • Everyday comfort : No-brainer too, and works silently over the network without any user intervention required. Good.
  • Easy data restoration : A full OS installation and length copy procedure is required. Bad.
  • Control on data : You own the NAS, but people who have access to your network can access it too. Average.
  • Robustness : The drive is in the same house as the computer, so major incidents (fires, etc…) are likely to destroy all copies of the data. Still, that’s pretty good protection already. Good.
Public clouds
  • Low financial cost : These services tend not to cost much per month, but the cumulative cost over several years tends to get higher than the price of regularly replaced external drives. Average.
  • Easy setup : If good software is provided, setting up cloudy backup is not too much of a hassle. Good.
  • Everyday comfort : Same as for setup. Software-dependent, but if the software is good it can be totally seamless (cf Apple’s iCloud for mobile devices). Good.
  • Easy data restoration : Requires a full OS install to get things operational again, and copy over the internet is usually at least an order of magnitude slower than from a local/networked drive. Very Bad.
  • Control on data : You handle your data to a random service provider which you don’t know well, and his servers can be compromised from the whole Internet. Very bad.
  • Robustness : Server owners can afford crazily paranoid setups where data is duplicated in several parts of the world that are far away from each other. With a good provider, the robustness is as close to perfect as it can get. Very good.
Robust server and thin clients
  • Low financial cost : For a home network, this option is quite a bit overkill and certainly jokingly expensive. Very bad.
  • Easy setup : It takes a motivated specialist and days of work to get something like this up and running, and even adding up new computers is complicated. Very bad.
  • Everyday comfort : Many modern OSs are designed with this use case in mind and make operating the thing daily seamless, as people who use such a solution at work can attest. Good.
  • Easy data restoration : Just take another computer in the home and log in, you get your session back. Good.
  • Control on data : It’s your server, but everyone on the network can see it. Average.
  • Robustness : The server is vulnerable to major incidents, but you’re not likely to physically damage it during normal use. However, if the server dies, everyone in the house loses access to its computer, unless heavier clients are used. Average.
  • Low financial cost : If you have several computers in your home and lots of disk space in them, which is common nowadays and will get even more common as we move towards individual computers like laptops, this will cost pretty much nothing. Good.
  • Easy setup : Operating system software can make it a breeze. Good.
  • Everyday comfort : Can happen silently in the background, like with a NAS. Good.
  • Easy data restoration : With proper OS support, we can envision making another computer “boot” on the saved backup image in order to continue work. Good.
  • Control on data : Same as above, but the other house computers can also be separately compromised when away from home. Though backup encryption can help against this, it does so at the cost of more complex procedures. Average to bad.
  • Robustness : Vulnerable to major incidents, but decentralized enough to be resistant to a horrible computer death. Good.


Here’s a quick comparative summary of the ratings given and explained before.

Method Cost Setup Use Resto. Ctrl Robust.
ExtHDD Avg Good Avg Bad Good Good
RAID Avg Bad Good Avg Good Avg
Partition Good Avg Good Bad Good Bad
NAS Bad Good Good Bad Avg Good
Cloud Avg Good Good VBad VBad VGood
ClientServ VBad VBad Good Good Avg Avg
CrossBack Good Good Good Good AvBad Good

As can be seen, some characteristics are easier to get than others. Nowadays, with modern software, ease of use is pretty much a given, and only hard drives get annoying due to the need to plug them in. Control on data, on the other hand, is more of a random thing, and quickly and easily getting a functional system back from a backup is still a lot to ask.

So far, my “cross backup” idea seems to perform extremely well, according to the criteria I’m considering as important. It combines all the advantages of a NAS with reduced cost through computer hardware reuse and improved ease of restoration (because the backup is not stored on a dumb storage device but on a working computer that can potentially make use of the backup file directly). It should be noted, though, that its greatness only exists given a pretty large set of assumptions, that I shall now recapitulate :

  • Computer evolution goes towards individual computers, like laptops and other mobile devices. As a consequence, having multiple computers of the same kind inside of a household, likely to be turned on at the same time, is probably to become the norm tomorrow.
  • We’ll continue to have gigantic-sized mass-storage devices inside of desktops and laptops, that no one but the extremely multimedia-savvy user will ever fill. Tablets will get them too eventually, and until then there’s no reason to worry as they are very far away from replacing a good laptop. Phones never will, their screen is just too small.
  • Computers inside of an average household, even the one of someone geeky enough to run my OS, will continue to run the same OS most of the time, making such cross-backup and easy restoration stuff simple to implement.

So, what do you think ? (100th post !)


One thought on “Backup : comparison and ideas

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s