Planning Storage Environments › Data Transfer Requirements › Alternate Data Path Considerations › Segment Your Network › Backup Scope

Backup Scope

After you have segmented your data, you can further reduce the required data transfer rate by reducing the scope of some backups. Typically, a relatively small percentage of your data changes from day to day. While these changes need to be saved, a full backup is usually unnecessary.

Example: Backup Scope

If you try to back up everything daily and only 10% of the data changes in the course of a day, you are spending 90% of your limited backup time storing data that is already backed up. When you include media consumption and wear and tear on your backup devices, this can be an expensive proposition.

You should consider backing up everything weekly, after 50% or more of your data has changed. You could then use the longer, weekend backup period for your longest storage operation. On a daily basis, you could back up the changes only. This would let you stay within the short, nightly backup window and would economize on media.

Arcserve Backup provides options for you to address this issue with the following types of backups.

• Full backups--stores everything, regardless of when the data last changed.
• Differential backups--stores files that have changed since the last full backup.
• Incremental backups--stores files that have changed since the last full or incremental backup.
• Synthetic full backups--for r16 or higher Windows Client Agents, synthesizes a previous full backup session and all incremental sessions to a full session without the need for previous incremental sessions.

Creating the right mix of full and partial backup operations is something of a balancing act. Ideally, you want each version of each piece of data backed up once. You want to minimize unnecessary duplication that consumes media and time. Therefore, you should keep the following considerations in mind:

• Full backups store all of your data at once. They produce a complete, coherent image of the data as it was at the time of the backup. They also store the backed up data together in a single, easily managed storage object. As a result, backup strategies that rely exclusively on full backups are usually inefficient because the relative percentage of new data in the overall data set is generally small. Full backups save too many files that are already adequately backed up by a previous storage operation.

  In exceptional situations, however, where the bulk of an organization’s data changes substantially over short periods, a plan that relies on full backups exclusively may be the best choice. Because, in this case, most of the data is fresh at any given time, the full backup may actually be less prone to needless duplication than a mix of full and partial storage operations.

• Incremental and differential backups let you avoid network congestion and excessive media consumption. They better fit your existing hardware and bandwidth constraints and mesh better with your users’ working hours. Incremental and differential backups are faster than full backups. If you do several of them between full backups, many files are still backed up more than once, because the differential backup backs up all files that have changed since the last full backup. This redundancy means that you can restore quickly, because all the data you need for a full recovery is stored in, at most, two data sets (the full and the last incremental).

  Incremental and differential backups are only economical when the volume of changes is small compared to the volume of the data set as a whole. When this is the case, you can store changes on a small amount of media that is rewritten frequently.

• For r16 or higher Windows Client Agents only, synthetic full backups also let you avoid network congestion and excessive media consumption. Synthetic backups are faster than full backups. After you execute the first real full backup (parent), you schedule incremental and synthetic full backups as needed. The synthetic full backup takes the first full backup and all the following incremental sessions and synthesizes them into one synthetic full session. If you need to restore files, only the synthetic full backup is needed since the last full session and all incremental sessions are combined. This redundancy means that you can restore quickly, because all the data you need for a full recovery is stored in one data set (the last synthetic full backup).