Backups: a Case for Redundancy

Tom Crouser November 10, 2013 Comments Off on Backups: a Case for Redundancy

A question commonly asked is what type of backup solution to select? Should you have a local backup, a cloud based backup or a hard drive based RAID (redundant array of independent disks) system? Well, the short answer is all of the above.

Redundancy: suspenders and a belt

Redundancy: suspenders and a belt

Each type of backup has its own strengths and weaknesses. Local backups are great for fast access to files, but what about if your shop burns down? Remote (cloud) backups skirt this issue by not being physically located at the site in case of a fire, but what about downloading ten terabytes of restoration data from a remote server. Exactly how many days can you afford to close your shop waiting to get back up and running?

Remember, it’s not just the potential of losing your data; it’s also the potential of losing time dealing with the restoration of that data. This is why we always recommend starting with a RAID system at the central file host.  A RAID system is two or more hard drives working in parallel: writing, reading and accessing the exact same data at the same exact same time. You commonly find these features when shopping for a server; however, regardless of where your central file repository is located on your network, it needs to be protected by a RAID. If one drives suffers a hardware malfunction, the next drive takes over sole operations with no restoration time required. Of course, you would want to replace the malfunctioning hard drive as soon as possible, because in a simple two drive RAID, you wouldn’t have any immediate redundancy once one drive dies. That could lead one to consider purchasing a three drive RAID from the start or even going with a four drive RAID.

Next, we add a local backup system to this central file depository, usually an external hard drive using the automated backup feature of your server or a third party software solution.  Usually, these drives automate copying files on a daily basis to the backup drive based on changes to the files themselves. On the Mac side, Apple’s Time Machine serves as a great local backup for Macs and there are plenty of software solutions that function in the exact same way for PCs. These local backups are great to add to individual computers on your network. You can also direct your local machines to back up critical files to the network file depository, which will provide the added benefits of the data residing on a server that is protected by a RAID and an external backup drive.

Finally, we add a cloud backup service. The configuration of these services is very important, as you want to limit the upload speed of the service or completely prevent it from operating during normal business hours. Nothing will crush your bandwidth faster than trying to upload several terabytes of data to a backup server. The first time you install a cloud backup system, it can take a very long time to get everything backed up initially (your upload speed is almost always slower than your download speed and I’ve seen it take weeks before the initial backup is complete). It depends on your upload bandwidth and the amount of data to be backed up. Of course, once the initial backup is finished, only the modified files get changed. So, the system will become faster over time, however – you should still limit it during normal business hours to prevent it from clogging up your normal Internet traffic.

Each of these weaknesses exposes a truth, and that is that backup redundancy is the best method for maintaining critical data. Remember, not all of your data is critical but the most critical data should have the most amount of redundancy. No backup solution is “set it and forget it”. Even the most automated backups need to be checked at least once a month to determine whether they are working. Countless times I’ve fielded calls from companies that had created automated backups over a year before, but the backup itself stopped functioning a week after it was setup. They only discovered this when they attempted to restore data. If you wait until you need to restore from a backup before you check to see if your backup is working, then you should assume you don’t have a working backup.

Taking it even further, let’s say you do have a complete backup set. If you can’t properly restore your data from a backup solution, then the backup solution is completely useless. Forget natural disasters, there is simply no amount of cloud backups that can compensate for your server’s hard drive crashing. Without an operating system to boot, how do you propose your server is going to get online to download the restoration data? Where did you put those server operating system installation discs, anyway? On the other hand, in the case of a fire that burns your office to the ground; you better have a backup that is located offsite. You can always buy replacement computers, but you can’t buy replacement data. What about a simple hard drive failure on the network’s central file depository? How well do you think your office computers would work without access to your server? Well, you better be protecting your server or central file depository with a RAID or one hard drive crash could bring the entire shop to a standstill.

As you can see no single solution will cover all of the disaster scenarios that your business faces. However, it is possible to cover most scenarios through the use of multiple backup methods to create a redundant system. Because, if there is one thing to take away from this is that redundancy works. A redundant RAID on your file server works against most hard drive hardware failures that would bring your server down. Local redundant server backups to external storage works for quick, easy and immediate access to your data at any time. Redundant cloud based backups work to protect critical data offsite in secure servers far away from local natural disasters.

The point is that the combination of these systems creates a redundancy backup network that keeps your doors open and your critical data protected with the least amount of downtime possible.  We can’t eliminate risk, but we can reduce the probability.

Redundancy works.

Comments are closed.