Backup every 30 min.

[Murigar]

I have a requirement to backup 1 server every 30 min.

I would like to do a Reverse Incremental schedule for every 30 min, 7 days worth. (336 recovery points)
Would 1 active full a week be enough?

To get longer retention, would it be advised to create a second job
With daily Forward Incremental for 90 says, Synth weekly, and Active monthly?

[Shestakov]

Hello,
1 Active Full will be enough in terms of backups` integrity, but be aware that if you try to restore from one of the first restore points of 336-long chain, it can take a lot of time.

In general 2nd job plan seems sane.
Could you specify what kind of longer retention are you trying to achieve? Maybe, you can just prolong the retention policy of the 1st job, splitting the chain to shorter intervals. Thanks.

[dellock6]

Another suggestion would be to use forever forward incremental instead of reverse incremental, thanks the better I/O profile it has chances are you can complete the backup every 30 minutes easier than by using reversed incremental, while having the same space savings.

[veremin]

And for longer retention you'd better put backup copy job with GFS mode into use. Thanks.

[Murigar]

To add a little bit of clarity. For this server in question. I do not necessarily need to be able to go back in time at the half hour mark. I just need a copy of changed data in the last half hour.

In reality I suppose a Forward Incremental 30 Min with 50 points, And a second job of 24 hour with 14 points would do what I want also.

I am certainly interested in Forward Incremental but I have concerns about completing a full restore which would require all 336 of the chain, is 50 more reasonable?

[dellock6]

that amount of restore point will surely slow down the start of any restore operation, regardless the method used. 50 point is obviously going to give you better performances. You can plan to pause replication during some off-hours to start another backup of the same VM that has longer retention but is done only daily.

[Shestakov]

Or, if off-hours are less critical, but still VMs should be backed up, you can schedule one job to run only in office hours with 30min. interval and second job to run only in off-hours time with, say, 2-hours interval.

[Murigar]

Or, if off-hours are less critical, but still VMs should be backed up, you can schedule one job to run only in office hours with 30min. interval and second job to run only in off-hours time with, say, 2-hours interval.

You could also accomplish this by running every "30 minutes" and setting some of the off hours to "Denied"?

[Shestakov]

Correct. We are talking about backup window.

[Murigar]

I have been successfully running this job for a few months but have a new issue I would like your input on.

The job is running a forward Incremental forever with 55 restore points.
I have a backup copy running daily to local storage with 90 restore points.

I now need to setup another backup copy to my DR site utilizing WAN Acceleration.
My problem is that on occasion I have had failures because the half hour file is locked. It can take quite a bit of time for this backup copy to occur.

Would it be wise to change this job to a traditional full/inc backup schedule with full/synth?

[foggy]

How are those backup copy jobs scheduled? What if you arrange them in time so that they do not try to read the same backup files simultaneously?

Also, if you're on v8, backup copy job shouldn't fail in case source backup files are locked, but just suspend and continue once the files are available.

[Murigar]

Primary Forward Incremental - Every half hour.
Backup Copy 1 (Local) - Every Day - 7 PM.
Backup Copy 2 (DR) - Every Day - 12 AM.

I am on version 8.

What do you feel is a reasonable chain length?
300 GB Full - 100 GB Inc.

[Shestakov]

"Reasonable chain length" depends on the size of increments, hardware productivity, RTO significance in a case of restore.
In terms of data consistency, long chaining is admissible, but restore from the end of the long chain, can take a lot of time.
Here is a relevant discussion about chains` length, full backups etc.
Thanks.

[Murigar]

Primary Forward Incremental - Every half hour.
Backup Copy 1 (Local) - Every Day - 7 PM.
Backup Copy 2 (DR) - Every Day - 12 AM.

....

Is there any way for (Backup Copy 2) to be built off of (Backup Copy 1) instead of the the primary data?
This would solve all of my concerns.

[Shestakov]

Yes, it`s possible starting from v8.
While the Backup Copy job itself cannot be the source due to its continuously running nature with no stop event, you can point the “child” Backup Copy to a backup repository used by the “parent” Backup Copy job to store backups in, to achieve the desired result. Thanks.

[Murigar]

To ensure I understand correctly. If:

Primary Forward Incremental - Every half hour. - BR1
Backup Copy 1 (Local) - Every Day - 7 PM. - BR2
Backup Copy 2 (DR) - Every Day - 12 AM. - BR3

Backup Copy 1 is using BR1 as its "Source"
Backup Copy 2 should be manually set to use BR2 as its "Source"

[Shestakov]

Yes, this configuration should work.