What's a Backup? (It's a Restore!)
Backing up your data is a critical process for any business, and as computers become more ubiquitous in our daily lives it is critical for many of us in our personal lives as well. Quite often this has never occurred to a person, whether personally or professionally, and at first it may seem complicated; But it's not.
A reliable and robust backup system really has two extremely important aspects:
- Your data must be replicated in a separate physical location, AND
- You must be able to restore from this off-site backup.
That's all there is to it! These two principles, applied together and tested on a regular basis, represent a backup system worthy of your highly critical data. Let's have a look at why these are the only critical aspects of a high-value backup system, and how easily they can be implemented in a fashion which meets your needs.
1. Locating Data Elsewhere (Off-Site)
Data redundancy systems such as RAID (Redundant Array of Independent Disks) can help you ensure high availability of the data stored within. RAID-1 (Mirroring) stores two identical copies of your data on two separate hard disks, so if either disk fails the data remains available. You can replace the failed disk and mirror the data from the working disk onto the new one, and quickly have two copies once again. RAID-5 (Parity) is a more complicated configuration which allows you to use more of the disk space while retaining the security that if one disk fails, its data can be reconstructed from the others.
But what if your redundant disk array catches fire and completely burns to the ground?
What if the basement where your file server is located floods?
What if there is a power surge affecting the data on all disks at once, since they are all running together?
Or what if two disks just so happen to fail in a short period of time, before the redundancy can be restored? These are all serious possibilities...
It is difficult for most of us to imagine these disasters. It's emotional. In my personal and professional experience, most people either don't think it will happen to them, or they simply put it off because it is difficult to think about.
The solution is so simple that it's almost too easy: Copy your data to a second physical location. Copy it on a regular basis (whatever that means for you and for your data). Think about the scenarios I outlined above; This works for all of them.
Off-site data storage can take many forms. Depending on the value you place on your data you should choose a system which you are comfortable with. Here are some examples of off-site data storage:
- Backup to tapes (to this day, tapes store high volumes of data in a compact space), and ship the tapes to a data storage vault.
- Send the data to another system across the Internet; This could be an encrypted data channel to another server under your ownership, or a third-party service such as Dropbox(tm).
- Copy your data to a USB key and give it to a friend.
There are a multitude of other options you could implement as well. Consider the cost of the options above; Clearly there is a big range between a high volume tape storage system with shipping and storage costs as compared to the last option which is extremely affordable. Obviously there are security issues to consider such as encrypting data before storage or transmission. Only give the USB key to a friend you trust...
All of these options would allow you to retrieve your critical data following complete loss of your live system (whether it is a server cluster or your personal laptop). For a business, a complete and documented Disaster Recovery plan is a good idea, and depending on the industry and data in question may be a requirement.
2. Restoring Data - Can You Do It?
A backup is only as good as its ability to be restored. If you simply assume your backups are working and up-to-date, you might find they are not, in which case you might as well never have bothered setting them up in the first place.
Let's think about some reasons a backup system could fail you, leaving your data unrecoverable and lost forever:
- The tape drive you were using for the past five years was misaligned, and every tape it ever created can only be read by that one singular tape drive. When your data centre burns down, you buy new equipment and retrieve your backups tapes from off-site storage. But wait! The new tape drive won't read any of your tapes! Your five years of backups are safe, but unusable.
- At some time a system configuration settings was changed, causing the backup process to be put on hold. No one happened to notice that the backup being copied elsewhere is never updated; The data is the same as it was last week. Without realizing it, the backups you've been sending off-site were actually the same file from five years ago, not the live production data from today.
- The free backup service you've been using goes out of business, or changed its Terms Of Service and affected your data retention. Perhaps a smaller limit was imposed and your backups are not full backups, truncated to the new service offering maximum storage size.
- Your CD/DVD optical media was left in a sunny room at your friend's house, and cannot be read.
These may seem unbelievable, even silly. But I have seen them all.
The good news is that all of these can be overcome with a relatively straightforward and easy practice: Test your backups; Ensure they can be restored on a regular basis.
If you test your restore process on a weekly basis, the most you risk losing is the last week's worth of work. You might choose to only do this monthly, quarterly, annually, or never. But weigh the effort it takes to review the backups and check up on the restore process against the effort it would take you to rebuild all that data from scratch, assuming that's even possible. Asking your clients to send you copies of all the invoices you've issued in the past year is unlikely to go over very well, while a short and well documented restore check can be very easy even for computer novices.
It's As Easy As One, Two (There Is No Three)
Clearly more can be done; You could send third and fourth copies to other locations, using a variety of hardware and software solutions to store data both online (ex: separate server or cloud storage) and offline (ex: tape, DVD, or USB disk), but by applying these two simple principles you will safeguard your data and know with decent certainty that it will be there when you need it.
Having no backup policy / procedure is very risky and will leave you stranded when a disaster (such as disk failure) occurs.
Having a second copy somewhere else, checked on regularly to ensure it can be restored, gives you a good assurance that you can get back up and running quickly. You might even sleep better at night.
Having additional mechanisms on top of this, or deploying multiple of the solutions suggested will improve your data security, but the greatest impact is achieved in just two steps.
These principles are based on my experiences professionally and personally over the course of many years. Trust me when I say I've lost many, many, many hard disk drives due to simple fatigue and wear and tear. I believe strongly that this simple two-step solution is beneficial for my friends and family as well as for critical business infrastructure. That said, if you have any feedback, agreements, disagreements, or stories of wonderful/disasterous backup policy, please use the comment form below.