World Backup Day reminds us all of just how important backups are. You don't get how important they are, perhaps, until you've experienced an outage that you can't recover from by any troubleshooting method. Backups are a pain but they are a necessary evil and can save you when things go bad. And things always go bad. This article helps you make a plan.
[ Reader also liked: 5 Linux backup and restore tips from the trenches ]
Select a backup method
I've found that the best backup method for me has always been to run a full backup at the starting point in a month and then run differential backups for the rest of the month. A starting point means at the beginning if you can manage to get full backups of everything on one day. If not, stagger your backups and then do differential backups for the next 30 days. It's a pretty simple idea.
Differential backups only back up what's changed since the last full backup. This is the most efficient method for a quick restore. Restore your full backup and then your latest differential and you're done. Differential backups also give you some idea of the volume of changed files you have which indirectly will give you some idea of future storage growth requirements. That's a topic for a whole other article, but just realize that storage growth is something you need to figure out.
Decide on backup storage locations
Keep a local copy of your latest full backup in two different locations. And keep your differential backups in two different locations as well. This will save you just in case of the destruction of one of the backups. A local backup will help you recover quickly from deleted files or a failed system.
Don't be "on the cheap" when performing backups or when storing backups on and off-site. Your backups are your company. They should be stored on reliable media and readily available for restore.
Set up disaster recovery options
Having hot spares of every critical system might not be financially feasible for you, but you can create virtual machines (VMs) to mimic your production environment and have them readily available to go online in case of a disaster.
Having an off-site location such as a remote data center that houses your disaster recovery environment is ideal but expensive. It all depends on how critical your operations are and how long your services can be offline before you lose revenue. Frugality is good except when risking business continuity to save a few dollars.
Test your plan
Always test your plan. First, you need to set up an annual mock failover plan meeting to discuss the plan of action in case your primary services are offline and what you'll do to restore them. This so-called tabletop scenario is good because everyone knows their place and when things really do go wrong, you have prepared.
On a smaller scale, you need to periodically restore a test file from backup to see how effective your backups are and how quickly you can restore files.
Revise as needed
No plan is good forever. Situations change. Business requirements change. Technology changes. Revise your backup strategy as often as needed to keep your backups clean and efficient for rapid restoration or failover.
Remember that everything works on paper but not necessarily in real life. Check your plan often and revise it as needed.
World Backup Day reminds us that we are vulnerable to failure and that backups are the best method to lower the risk. Planning, executing, learning, and revising are the four tasks required when working on a backup strategy.