Synthectic Full backup

[Gale_Bishop]

How do the synthectic full backup will work?

What is the design of synthectic full backup.,

Will it be scheduled automatically once aftre reaching a specific criteria? if yes, what is the creiteria?

[RahulG]

HI there refer to the following document
http://support.veritas.com/docs/270015 
The document is for version 10d but the concept remains the same

A Synthetic Full Backup is an operation that combines data from prior Full and Incremental backups to produce a dataset that is indistinguishable from a Full Backup created at the time of the last Incremental job.

The concept of Synthetic Full Backup job is to perform Full backup of the system data only once and then perform subsequent incremental backup jobs. The backup sets obtained from the Incremental jobs and the Full backup jobs are then combined to produce a backup set that will replicate the backup set if a Full backup job was to perform on the last day the Incremental job was fired. Synthetic backup is an add-on feature which is added when the Advanced Disk-based Backup option is added.

A policy for Synthetic backup job will comprise of Full (baseline) job, Incremental job, and a Synthetic Full Backup.
Due to the interdependent nature of jobs and to exercise efficient control of data, Synthetic Full backup jobs must always be performed within a policy.

Note the following

• Baseline: Scheduled to run once a month (or more frequently) to set a baseline for further backups.
• Incremental: Scheduled to run frequently as normal incremental would.
• Synthetic Full: Scheduled to run once a week.

Hope this helps

[RahulG]

Also go through the Best practice guide http://support.veritas.com/docs/285747

[Colin_Weaver]

Synthetic Backups work as follows

Run A Baseline Full to start the process.
Run one or more incremental backups
Run a synthetic backup which takes the basline full and any incrementals since and makes them into a new synthetic full
Run more incrementals
Run another synthetic backup which takes the presious synthetic full and combines it with the incrementals since and makes a new synthetic (full)
Repeat...

To make each synthetic the presvious 'full' and all the 'incremental' backup sets must be available and online - which usually means it is worth keeping the whole concept running to Backup to Disk Storage and then add a duplication phase to get the data to tape (if required)

You do need an ADBO license to use synthetic backups - however other than this it is just a policy based job containing templates with the gather synthetic info enabled


A word of warning though a large number of changing files could cause the gathering info for synthetic phase of the Incremental jobs to take a long time - which in extreme circumstances can actually end up being longer than running a traditional full backup.

[Bret_Brutlag]

I pretty much have a handle on how to setup synthetic backups and the general concept. But I have a couple questions about disk space usage and restores before I commit to the plan.

Assuming Synthetic backups are in place with a 1x per week full and incrementals the rest of the week, dup to tape weekly.

1. how do restores work? If I have a user that tell me that they discovered a file needs restored from a state  3-4 days ago is that possible? it is possible that it was changed 1 day ago so it would be included in a recent incremental backup.

2. Currently I am doing weekly full, daily differential , weekly dup to tape. I am able to keep 6 days of B2D in local storage.  B2D media has a 6 day retention. File are 200gb files (35 daily jobs). Would Synthetic configured as described be more efficient in the storage part.

Thanks