For any business with data to protect, Rule #1: Back Up Your Data. Rule #1a is Verify Viability for obvious reasons, but first you need something to work with. So, what’s the best backup strategy for you? It depends on how much data you have and how much of it can you afford to lose.
Full versus Incremental and Differential
A full backup copies all files in a given location. An incremental backup copies only the files that have changed in a given location since the last backup. The backup software determines this by making use of the archive bit. When a file is backed up by either a full or incremental job, its archive bit is cleared. When a file is later modified by a user or the system, the archive bit is set again, thereby flagging it for the backup process to copy again during the next job run. A differential backup is very similar to an incremental backup, except that it copies only the files that have changed since the last full backup. The differential backup accomplishes this by leaving the archive bit set.
Full backups are generally not practical to do on a nightly basis because of the amount of time they take to complete. You still need to do them though because they are your starting point in the event of a disaster that requires a full recovery. Standard practice is to combine two of the options by performing a full backup once a week (generally on the weekend, when you have the longest stretch of time without users in the system) and nightly incremental or differential backups on the other days. The advantage of differential backups is that, in the event of a disaster, you will only need your last full backup and your last differential backup to recover. The disadvantage when compared to incremental backups is that each day that passes since the last full backup takes longer to back up the changes. Incremental backups are more efficient during the backup process, whereas differential backups are more efficient during the restore process.
Whichever combination you choose, nightly backups will satisfy most organizations, ensuring that you will not lose more than 24 hours’ worth of changes. If 24 hours is too large of a risk window for you, another option is to use ghost imaging, which synchronizes the files between two network locations at regular intervals (e.g. every 15 minutes).
Disk versus Tape
Ghost imaging is, by nature, a disk to disk backup method. Disk to disk is generally faster than backing up to tape, but also potentially more risky because you can’t move the copy off-site. While disk to disk backups across a WAN are an option, they can be slower than disk to tape depending on your bandwidth capabilities and run the risk of failing if the connection between the two sites is lost. If your building gets flattened and you need to set up shop in a new location, you are going to need those backups. If the tornado that just ripped through your building deposited them in a nearby (or not so nearby) tree, they’re probably not going to be worth much. Rule #2: Store Backup Copies Off-Site. Depending on the types of natural disasters prevalent to your area, storing them at least 10 miles away is a good place to start.
Another important consideration when backing up to tape is how many times you can re-use your tapes. While they are built to be written to multiple times, after a while doing so will cause degradation and threaten the integrity of the data. The most common way to manage this is by using the Grandfather-Father-Son (GFS) backup strategy.
The GFS backup strategy is a method of maintaining daily, weekly, and monthly backups based on a five or seven day weekly schedule. A full backup is performed once a week. Incremental or differential backups are performed on the other days. The daily incremental or differential backups are known as the Son. The Father is the last full backup in the week (the weekly backup) and the Grandfather is the last full backup of the month (the monthly backup).
To maximize your backup and restore capabilities, you can re-use daily tapes weekly. Weekly tapes can be overwritten after five weeks. Monthly tapes should be saved for a year and most definitely taken off site for storage. One of the great benefits of the GFS strategy is that it assists with determining the minimum number of tapes required and establishing a consistent interval at which to rotate and retire them. The following table shows an example GFS implementation on a five day schedule over two months where the entire backup can be stored on one tape.
|Month 1 (Full-W=Weekly backup, Full-M=Monthly backup)|
Using the above rotation you can calculate that it will take a total of 21 tapes to perform this backup strategy for one year. There are four daily tapes (Sons) that are reused weekly, five weekly tapes (Fathers) that are reused after the fifth full weekly backup is complete, and 12 monthly tapes (Grandfathers) which are the last full backups of the month and are taken offsite. The daily tapes clearly get the most use and should therefore be replaced more frequently. Follow the manufacturer’s recommendation for the type of tapes you are using, but a general rule of thumb is to plan to replace these tapes every year.
While all of this strategy is great, it isn’t worth much if you have too much data to back up in too small of a time window. If you can’t perform a full backup in less than 12 hours, it’s time to look at splitting your data up into multiple disk locations with independent backup processes so that you can run simultaneous backup jobs on each one. Why not 24 hours if the full backup is being performed on the weekend anyway? Because, if your backup fails, you don’t want to have to wait a whole week to try again.
Do you have a favorite backup tip or a question about backup strategies? Please share it in the comments!