Andrew Hudson - Fedora™ Unleashed, 2008 edition

All these accidents and disasters offer important reasons for having a good backup strategy; however, the most frequent cause of data loss is human error. Who among us has not overwritten a new file with an older version or unintentionally deleted a needed file? This applies not only to data files, but also to configuration files and binaries. Among users perusing the mail lists or the Usenet newsgroup postings, stories about deleting entire directories such as /home, /usr, or /libseem all too common. Incorrectly changing a configuration file and not saving the original in case it has to be restored (which it does more often than not because the person reconfigured it incorrectly) is another common error.

To make a backup of a configuration file you are about to edit, use the cpcommand:

$ cp filename filename.original

And to restore it:

$ cp filename.original filename

Never edit or move the *.originalfile, or the original copy will be lost.

Proper backups can help you recover from these problems with a minimum of hassle, but you have to put in the effort to keep backups current, verify their intactness, and practice restoring the data in different disaster scenarios.

By now you realize that some kind of plan is needed to safeguard your data, and, like most people, you are overwhelmed by the prospect. Entire books, as well as countless articles and whitepapers, have been written on the subject of backing up and restoring data. What makes the topic so complex is that each solution is truly individual.

Yet the proper approach to making the decision is very straightforward. You start the process by asking

► What data must be safeguarded?

► How often does the data change?

The answers to these two questions determine how important the data is, determine the volume of the data, and determine the frequency of the backups. This in turn determines the backup medium. Only then can the software that accommodates all these considerations be selected. (You learn about choosing backup software, hardware, and media later in this chapter.)

Available resources are another important consideration when selecting a backup strategy. Backups require time, money, and personnel. Begin your planning activities by determining what limitations you face for all these resources. Then construct your plan to fit those limitations, or be prepared to justify the need for more resources with a careful assessment of both backup needs and costs.

If you are not willing or capable of assessing your backup needs and choosing a backup solution, a legion of consultants, hardware vendors, and software vendors would love to assist you. The best way to choose one in your area is to query other Unix and Linux system administrators (located through user groups, discussion groups, or mail lists) that are willing to share their experiences and make recommendations. If you cannot get a referral, ask the consultant for references and check them out.

Many people also fail to consider the element of time when formulating their plan. Some backup devices are faster than others, and some recovery methods are faster than others. You need to consider that when making choices.

To formulate your backup plan, you need to determine the frequency of backups. The necessary frequency of backups should be determined by how quickly the important data on your system changes. On a home system, most files never change, a few change daily, and some change weekly. No elaborate strategy needs to be created to deal with that. A good strategy for home use is to back up (to any kind of removable media) critical data frequently and back up configuration and other files weekly.

At the enterprise level on a larger system with multiple users, a different approach is called for. Some critical data is changing constantly, and it could be expensive to re-create; this typically involves elaborate and expensive solutions. Most of us exist some where in between these extremes. Assess your system and its use to determine where you fall in this spectrum.

Backup schemes and hardware can be elaborate or simple, but they all require a workable plan and faithful follow-through. Even the best backup plan is useless if the process is not carried out, data is not verified, and data restoration is not practiced on a regular basis. Whatever backup scheme you choose, be sure to incorporate in it these three principles:

► Have a plan— Design a plan that is right for your needs and have equipment appropriate to the task. This involves assessing all the factors that affect the data you are backing up. We will get into more detail later in the chapter.

► Follow the plan— Faithfully complete each part of your backup strategy, and then verify the data stored in the backups. Backups with corrupt data are of no use to anyone. Even backup operations can go wrong.

► Practice your skills— Practice restoring data from your backup systems from time to time, so when disaster strikes, you are ready (and able) to benefit from the strength of your backup plan. (For more about restoring data, see the "Using Backup Software" section later in this chapter.) Keep in mind that it is entirely possible that the flaws in your backup plan will become apparent only when you try restoring!

You have to create your own best-backup plan, but here are some building blocks that should be incorporated into the foundation of any sound backup program:

► Maintain more than one copy of critical data.

► Label the backups.

► Store the backups in a climate-controlled and secure area.

► Use secure, offsite storage of critical data. Many companies choose bank vaults for their offsite storage, and this is highly recommended.

► Establish a backup policy that makes sense and can be followed religiously. Try to back up your data when the system is consistent (that is, no data is being written), which is usually overnight.

► Keep track of who has access to your backup media, and keep the total number of people as low as possible. If you can, allow only trusted personnel near your backups.

► Routinely verify backups and practice restoring data from them.

► Routinely inspect backup media for defects and regularly replace them (after destroying the data on them, if it is sensitive).

Now that you are convinced you need backups, you need a strategy. It is difficult to be specific about an ideal strategy because each user or administrator's strategy will be highly individualized, but here are a few general examples:

► Home user— At home, the user has the Fedora installation DVD that takes an hour or so to reinstall, so the time issue is not a major concern. The home user should back up any configuration files that have been altered, keep an archive of any files that have been downloaded, and keep an archive of any data files created while using any applications. Unless the home user has a special project in which constant backups are useful, a weekly backup is adequate. The home user is likely to use a DVD-RW drive or other removable media for backups.

► Small office— Many small offices tend to use the same strategy as home users, but are more likely to back up critical data daily and use manually changed tape drives. If they have a tape drive with adequate storage, they will likely have a full system backup as well because restoring from the tape is quicker than reinstalling from the CDs. They also might be using CD-RW or DVD writers for backups. Although they will use scripts to automate backups, most of it is probably done by hand.