Being a backup administrator it is important to make a backup plan to protect your environment.
There is a huge difference between Full, Incremental, Differential and Duplicate backups. In order to optimize the backup operation and save storage space, it is very importance to create a backup method depending upon the criticality of the data which would help you to recover your data in the shortest possible period during any event of disaster.
How to design an effective Backup Plan
- Type of data and its criticality
- How frequent the data changes everyday so that you can run backups accordingly
- How much is the base backup and changed data hourly/daily/weekly/monthly and how much storage space you may need.
- How to keep a track of growing data and storage space
- In case of tape storage, how many tapes you may need and rotate tapes daily/weekly/monthly/yearly etc.
Besides the above point you need to design your backup schedule that is also very important task to optimize the backups. Now scheduling the backup jobs again depends upon the overall data change across your environment including all departments. Also time taken to rebuild a particular server or database in case of hardware failure.
For example if you have a financial database which is very critical data and it changes hourly, then you may consider to run full backup daily and incremental backup every two hours / four hours which will ensure that every change is backed up. You should also consider running differential backup every weekend.
Lets discuss the types of backups.
Full backup is considered a baseline backup which creates a reading points for other type of backups like Incremental and Differential. Full Backup will backup the whole data which is selected for a a backup of a server or source device each time it runs. It will not check or reset the archive bit or anything.
Any file whether is modified or not, each time you run a Full Backup, it backs up everything. It includes backup to a tape library, tape storage, disk storage and cloud storage where it will backup the full data in every Full backup and does not save any space, in fact Full backup occupy a lot space in the storage.
Besides that it would occupy same amount of data as per the total data size. Full backup will not do any de-duplication or avoid duplicate data. However you run full backup to a de-duplication storage then it would not backup same data regardless of type of backup and would backup only unique data to the dedup storage.
Incremental backup will backup data which is changed since last full or incremental backup. It will back up only block level changes since the last full and incremental backup. But the differential backup will backup all data changed since the last full backup.
Note: It is advisable to use same storage device for full and incremental backups which is accessible to back exec server.
Backup window of incremental backup will be shorter since it would backup only the changes since previous incremental or full backup.
-Full backup runs on Sun which is the baseline backup.
-Incremental backups run on Mon, Tue, Wed, Thu, Fri and Sat.
-Now the incremental backup which runs on Mon will backup only changed data since full backup of Sun
-Tues incremental will backup data changed since Mon incremental
-Wed incremental will backup data changed since Tues incremental
-Thu incremental will changed data since Wed incremental
-Fri incremental will backup changed data since Thu incremental
-Sat incremental will backup changed data since Fri incremental
Again on Sun a full backup will run and same cycle will follow
Now if there is hardware failure occurred before next full backup runs, then you will need a last completed full backup and all the completed incremental backups to restore the complete data.
Lets check the above example where full backup runs on Sun and incremental runs Mon – Sat. Now if the backed up server crashes on Sat then you would need last Sun full backup and all the incremental backups which ran from Mon-Sat so that server can be restore till Sat. If you restore full backup and lets say incremental till Wed only, then server will not have all the data. The data of Thu, Fri and Sat would be missing.
The differential backups will backup all the data which was originally selected in the full backup, changed since the last full backup. And the incremental backup will backup only the data changed since the last full or incremental backup. It depends upon you how you want Backup Exec to detect change of data which can be configured by modified time / archive bit / Backup exec catalogs.
Also, it is advisable to run Full and Differential backups in same storage device accessible to backup server.
Note: Backup selection list should not be changed without completing a full backup cycle. In case any backup selection list is changed then a Full backup should run first then the following incremental or differential or both should be run.
Differential backups will consume a lot of space since it will have all the data changed of previous full backup. However, it helps to restore complete server data faster as you need only a full backup and last differential backup which results in fewer media files in the storage device lesser changes of restore job failure.
Duplicate backup is a copy backup of any backup where duplicate backup is configured. You can run duplicate backup backup Full, Incremental and Differential backup. For a duplicate backup you will need two type of storage as duplicate is configured between two storage device.
If you have only one type of storage then duplicate job is not possible. However you can duplicate backup between two similar storage devices like Disk to Disk but it is not considered a good strategy.
Normally, you can run duplicate backup from Disk to Tape or vise-versa. You can also run Duplicate backup from Disk to a Dedup storage or Tape to Dedup which is the best way to save storage space as well. You can also run duplicate backup from Dedup storage to a disk to tape which will also save storage space as dedup storage will backup only unique data and avoid duplicate data.