As you may know, I participate in several on-line IT discussion forums. Every few weeks, a new member will post a question like “I have an xyz server, what should I use for backups?”. Seemingly helpful forum members quickly jump in and start throwing out vendor names, do-it-yourself solutions, discussions about NAS versus SAN and disk-to-disk versus disk-to-tape, and so on.
It makes me want to SCREAM? Why? Because …
What you use for backup is irrelevant unless you know what you are backing up, why, and how quickly you need to restore!
Said another way, before you pick ANY backup solution, you should know:
- What you need to restore
- Why you expect you will need to restore or recover it
- How quickly you will need to restore or recover it
Note that the answers you provide may vary for the different types of data. For example, you may be able to live without your accounting system for two days, so long as you can ship orders within 4 hours. You may need current project files immediately, but could wait a week for projects completed more than a year ago.
Understanding your “Why”
When considering why you might need to restore or recovery information or systems, think of the full spectrum of activities that can go wrong. As an informal set of definitions:
- Restore operations are usually performed on individual files or small sets of data, often resulting from accidental overwrites, deletions, or component (disk) failures.
- Recovery operations are usually geared for large data loss, such as a drive array failure or server loss due to a disaster.
The key difference, restore operations bring back select data from a specific point in time while recovery operations bring back entire systems or data environments.
For example, you might restore email messages accidentally deleted from a user’s account on an MS Exchange server. If the disk array dies, however, you would need to recover the entire mailbox store for the server.
Backup/Recovery protects you from disaster; Backup/Restore protects you from component failures and user errors (or intentional misconduct).
How you backup for recovery will often differ than how you backup for restore.
Backup solutions that efficiently restore data, are not optimized for recovery. Most backup solutions designed for fast recovery, such as image snapshots, lack the ability to restore individual elements. For the Exchange server, above, we would recommend running two backups — one designed for recovery and one for individual mailbox and message restores.
Additionally, backups for restore generally give you more retention points than backups for recovery. Being able to select a specific time or version of data is a key feature for backup/restore solutions.
Understanding you “How Quickly”
How quickly you need your data depends on the data and your business. Keep in mind that you do not need all of your data all at once. Generally speaking, however, when you need to restore an active file or two, you want to be able to do this quickly. While you want quick recovery as well, you are more likely to be bound by factors beyond your backup/recovery solution, such as purchasing new hardware or moving to temporary office space.
Focus first on how quickly you need to Return To Operations. Your RTO will drive your selection and investment in backup/recovery solutions. Once you have your RTO, identity the critical data and systems you need to get your business up and running. Your RTO will be shorter than your window for full recovery, and includes only the critical subset you need to get up and running.
The shorter your RTO, the more expensive the solution. A realistic RTO will prevent you from over-buying.
First Steps First
By first understanding your requirements — the what, why, and how fast — of your restore and recovery needs, you can select backup solutions that accurately match your needs and effectively protect your data and your business. By defining your needs, your solution will be relevant and your investment well-made.