Most people know that inevitably all computer systems experience some sort of failure. If you deal with any business, client, or critical data of any kind you are also probably aware that backups of that data are the key to not losing the information forever in the event of a serious enough system failure. The question is what backup method is right for you?
Not all backup strategies are equal and to be fair there is no one size fits all solution for backing up your data. Typically there are two categories for backing up data: Disaster Recovery, and One Time Restores. Your needs, the amount of time you can be without your data, and other factors will help govern what backup strategy is right for you.
What Category do I fall into?
While it’s always best to talk to an expert about your backup needs, you can try to gauge what is right for your situation; the more you know the more you can rest assured what is in place will protect your data sufficiently. Below I will describe a bit about each category.
Disaster Recovery is the big one, and there’s virtually no good reason to not have at least one full backup of all your systems and data. What we are talking about here is a complete system failure; typically a Hard Drive failure, an update or change that causes your operating system to not load, or something that causes your data to become in-accessible. There are many caveats, types of failures, and a wide range of solutions that suite different needs. You can try to mitigate a failure by setting up highly available systems. Technologies exist that allow you to use multiple hard drives to create a Redundant Array of Inexpensive Disks (RAID) so if one fails your data is not lost and the failed drive can be replaced safely. There is also technology to replicate your data to a backup server and experience virtually no downtime. All of this is called High Availability and will be talked about in a future article.
Regardless of having a backup server or not, the equipment that failed needs to be brought back to an operational state as soon as possible. The quickest way to do this is to restore a complete image or copy of the hard drive from it’s most recent working state. A complete restore like this on a new hard drive, or over top of an existing hard drive with corrupt data will save time and effort as it should bring the computer to a working and bootable state. Depending on your backup strategy this may be sufficient or you can combine this with the restore of a One Time Restore (explained further down) to provide the best of both worlds.
The main point of this type of backup is it’s all encompassing, and is typically an image or low level dump of the hard drive as opposed to individual files and folders. It will backup up pretty much everything and restore it like a snapshot in time. At the very minimum a full backup like this should be taken at the install of any new system, and updated at least monthly, if not more frequently as changes to installed programs and security updates will make the original backup obsolete. If this is your only method of backing up your critical files and data as well as the systems they are run on you will want to make sure this process occurs at least as frequently as you are willing to lose data. Generally a daily or weekly backup will suit most peoples needs.
Lots of programs exist to help you with this task and most modern operating systems even have built in utilities for this. In Windows you can utilize the built in Windows Server Backup, and in older versions NTBackup. For Linux and BSD based machines the built in Dump command is usually sufficient and can be combined with compression technology like BZip to create a smaller backup size. If you want a third party program that can help you out and provides a nice clean interface I would recommend checking out Backup Assist.
The amount of space you have to store your backups will help determine what your ideal backup strategy is and as you can imagine an entire image or backup of a hard drive can take up a lot of space. For this reason you are able to configure incremental and differential backups which will take a look at the last full backup and only copy the changes made since that time saving space but slightly increasing the time to restore your data.
One Time Restores
One Time Restores are different from Disaster Recovery backups in the sense that they are typically meant for restoring a few files here and there or a One Time Restore of your critical data. Normally these types of backups will be useful for Web Hosting, Office Documents and anything that has a high likelihood of needing to restore a few files here and there but not the whole server and operating system. For some situations it may be satisfactory to even have a few Full Disaster Recovery Backups (say once a month) and then more frequent backups of just the critical files on that server. This allows the entire operating system to be restored to a recent working state and then the critical data to be restored to a very recent date.
A good example of when this comes in handy is if an important document is being worked on and gets deleted, overwritten, or changed and the original is needed. It doesn’t make sense to spend all the time and effort trying to restore or extract that one file from your entire system backups; especially if the most recent one is from two weeks ago. Since you are not backing up the entire operating system and only the data you use and update on a regular basis (client files, website data, contracts, etc) the size tends to be smaller and there are a lot of different technologies that let you keep more frequent backups. In Windows the Volume Shadow Copy service is a great system that will take snapshots of your data at regular intervals. Programs like Backup Assist also perform similar tasks and in either case each subsequent snapshot simply logs the changes to your files and not the entire content again. Depending on how frequently you update your files a lot of different snapshots can be stored on a relatively small amount of disk space; it may not be unreasonable to take a snapshot 2 or 3 times a day and only ever lose a couple hours of work at the most.
All backup solutions are great in theory but the true test is performing a restore and making sure it works. While only a small point, it is a very important one. No matter how long your backup solution has been in place it should include regular tests to ensure that all the right stuff is being backed up and it actually restores. It is far better to pick one backup a month or every couple of weeks to test and make sure it’s working as planned as opposed to finding out there was an issue when you are trying to recover the data after a failure.
While I can only scratch on the surface of this topic as there is so much that varies for every person and businesses situation and so many different possibilities, it is clear that for most situations a combination of Full Disaster Recovery, and One Time Restore backups is the best possible solution. Things to be aware of are: your storage space (for backups), how frequently the data is updated (hourly, daily, weekly, monthly?), and the amount of time you can spend restoring an individual server or file/folder. Hopefully this information has helped you figure out your backup situation, and as always please be sure to speak with an expert to discuss your individual needs and concerns.