You are only as good as your last backup.

Image via Wikipedia

Those are commonly used words, but think about it for a second.  Is your backup timely, available and worthwhile?  When catastrophe strikes, what is on the backups you have?  Do you have off-site plans, so that in case your servers are destroyed, your business is not? Have you ever backed up and then restored onto a new machine? Have you implemented a process where you regularly test your backups to see if you can use them?

Backups are like a nuclear bomb; unless you have run a successful test, you don't have one.

   I am often called to work at company locations where the servers have dust collected over the last decade, tape media which has been used over the life of the machine without getting rotated out, and nothing off-site.  As I am looking in horror at the task ahead of me, the principals of these companies re telling me how critical the systems are to their business.  This is also true of many enterprises, where although backups abound, they have not been tested on a regular basis.
   This problem, like many technology solutions has no one right answer.  The answer depends upon the criticality of the data, your acceptable level of downtime, and the cost and complexity that you are willing to accept for the solution.  The rule of thumb for any organization, as long as they use computers and software, is to have one regular backup on site, and another one at a remote location.
   Local or onsite backups can either go to a disk array specifically for backups or to a tape, where data is backed up on a regular basis.  The first step in any backup is to do a full backup, which copies down everything as it is onto backup media.  This is not a once-and-done type of backup, and should be performed regularly.  Monthly, or quarterly full backups are acceptable, but doing an annual full backup, while fine for non-critical information, is not recommended for your most critical data.  Your accounting database, any operational software, and email data is usually the most sought after subsequent to a system crash and therefore is generally the most critical.
   After a full backup is performed, then differential or incremental backups can follow.  Differential backups will look at the data and see what has changed since the last backup, and then only back up the changes.  An incremental backup is roughly the same, except that when restoring you would first restore the full backup and then the incremental backups sequentially.  There are many other differences besides that, but I’m trying to keep this as a blog for the masses, not just for geeks.
   These backups also need to be done to a remote location, often a hosted backup service.  There are quite a few vendors out there for this, one of them has famous commercials where people put their laptops in the microwave or get hit by falling satellites.  Many larger organizations have this policy in place where there is a remote site that backs up the data and serves as a data repository or redundant data center.  The insurance policy that this buys for a company is worth every penny.  The onsite backup procedures are helpful when a file is lost, something gets corrupted or a server crashes.  But what if the physical location of the server area is inaccessible or destroyed?  Without a reliable offsite backup, your company may never recover.
   The key to backing up your data is to test a restoral of your data regularly.  Until you do that, you cannot be sure if everything is working correctly.  Depending on your company’s tolerance for risk, this could be done monthly, quarterly or at worst annually.  If this is done as a policy regularly, then good procedures about how to get systems back online will be easily developed, and in a time of crisis you will have assurance that at least the systems can come back up.

Related articles by Zemanta

• Online vs Offline Data Backup: Part 1 (brighthub.com)
• City of Ettlingen Chooses "Hassle-Free" Virtualized Backup and Restore from Vizioncore (eon.businesswire.com)