Recommended method for weekly offsite full tape

[davidkillingsworth]

Hi,

I have one customer who would like to have 12 weeks of weekly offsite tape backups in rotation. We have 12 tapes.

So far the following:
I have 2 VMs with a full backup size of about 250GB
I have 2 Backup respositories (d:\fullbackups, d:\dailybackups)
I have configured 1 storage pool.
I have 2 Backup Jobs
I have 1 Tape Jobs

Job Type: Backup
Job Name: FULLWEEKLY
The first job is run weekly on Wednesday evenings at 10PM.
Repository: d:\fullbackups
Reverse incremental
Restore points to keep on disk: 4 (does this have any impact on tapes?? If doing reverse incremental and this is 4, will 4 full copies be kept on disk?)
Configure Secondary destinations for this job (checked)
Secondary Target (Tape Job - Fullbackuptotapejobname with 12 tapes free)

Job Type: Daily
The second job is run daily at 8PM
Repository: d:\dailybackups
Incremental
Synthetic Full (Sundays)
Restore Points to keep on disk: 14

Job Type: Tape
Name: Fullbackuptotapejobname
Backup Files: FULLWEEKLY - type VMware backup
Media Pool: Media Pool 1 (12 tapes)
Process incremental backup files (NOT CHECKED)
Eject Media: checked
Export current media set once the job finishes: (NOT CHECKED)
Run this job: After FULLWEEKLY


Questions:
* Do I have this configured most efficiently? The goal is to be able to archive as much of the full VM images offsite as possible. 12 weeks would be great, but if we could extend that by having more than one full backup on each tape, even better.
* What happens when the 12 tapes fill up?
* Will the tapes be overwritten each time, or will they append the last backup files.
* Are there any other concerns or suggestions?

[veremin]

Do I have this configured most efficiently?

From my perspective, you've specified almost everything correctly. The only missing part is retention setting of the given media pool. If you want to rotate tape mediums on 12 week basis, then, you should set "12 weeks" in corresponding field ("Protect data for").

What happens when the 12 tapes fill up?

It depends on retention settings mentioned above. Either the tape will be overwritten, or the job will be sitting and waiting for a new appropriate tape.

Will the tapes be overwritten each time, or will they append the last backup files.

It depends on "Media Set" setting of the media pool. With "don't create new mediaset", VB&R will always append data to the medium. Otherwise, it will act in accordance with the specified settings.

Are there any other concerns or suggestions?

I'm wondering whether "source" (weekly) job performs full backup from time to time. The idea is that sometimes there might be more than one full backup in given location, and they will be copied by backup to tape job. Which, in its turn, is not inline with your needs and desires.

Thanks.

[Dima P.]

Hello,
Additionally, please note, that backup to tape job processes only VBK (full backups) and VIB files (forward increments). Reversed increments (VRB) are skipped form processing.

Do I have this configured most efficiently? The goal is to be able to archive as much of the full VM images offsite as possible. 12 weeks would be great, but if we could extend that by having more than one full back up on each tape, even better.

If I calculated correctly: 250 GB weekly full would generate 12 TB in a year.
Assuming that you are using LTO 5 media (?), which is about 1.5 TB: 12 tapes x 1,5 TB = 18 TB
So, you would have 4 TB free left on tape media, which could be filled with additional portion of backups.

What happens when the 12 tapes fill up? Will the tapes be overwritten each time, or will they append the last backup files.

Let’s say - 1,5 TB media could keep 6 backup sets of your 250 GB weekly fulls. VBR would overwrite such tape when the last backup set would be expired. So in case you set up media pool retention for 1 year, such tape would be reused in 1 year + 6 weeks.

Are there any other concerns or suggestions?

Check what compression level you use on backup to disk jobs, in case it is default or higher - disable hardware compression on tape job, it would save your tape library processing time.

[davidkillingsworth]

I'm wondering whether "source" (weekly) job performs full backup from time to time. The idea is that sometimes there might be more than one full backup in given location, and they will be copied by backup to tape job. Which, in its turn, is not inline with your needs and desires.

No, it does not. I was wondering whether the reverse incremental did this automatically, or if we really need to check the "full backup" If we need to check the full backup, should we set it to do this every Wednesday? I thought that Reverse Incremental was a full backup every time.

I want to have a fully functioning tape that I can use to restore the backups in case of disaster. This is the purpose for using tapes in the first place. We can't use Cloud based backup because their Internet connection is very slow.

[davidkillingsworth]

So which should I be using - Reverse Incremental, Incremental, Incremental with Synthetic full, Incremental with Transform previous full with backup chains into rollback, any of the previous options with perform active full backup? The options are somewhat confusing if we just want to get the full backup to tape and offsite.

Do I have this configured most efficiently? The goal is to be able to archive as much of the full VM images offsite as possible. 12 weeks would be great, but if we could extend that by having more than one full back up on each tape, even better.
If I calculated correctly: 250 GB weekly full would generate 12 TB in a year.
Assuming that you are using LTO 5 media (?), which is about 1.5 TB: 12 tapes x 1,5 TB = 18 TB
So, you would have 4 TB free left on tape media, which could be filled with additional portion of backups.

What happens when the 12 tapes fill up? Will the tapes be overwritten each time, or will they append the last backup files.
Let’s say - 1,5 TB media could keep 6 backup sets of your 250 GB weekly fulls. VBR would overwrite such tape when the last backup set would be expired. So in case you set up media pool retention for 1 year, such tape would be reused in 1 year + 6 weeks.

Sorry, I should have specified LTO4.

So if I understand correctly, I need to set an retention on the pool. I have set "Do not protect data (cyclically overwrite tapes as required)." I think that will meet my needs. The tapes will be appended to as long as there is space, but will be overwritten when there is not enough space.

[veremin]

No, it does not. I was wondering whether the reverse incremental did this automatically, or if we really need to check the "full backup" If we need to check the full backup, should we set it to do this every Wednesday?

In case of reversed incremental mode, the latest restore point is full backup that keeps updated during every job run - the changes are literally injected into full backup. Meanwhile, the corresponding option ("Run active full backup") refers to the creation of independent full restore point. Such restore point is made from scratch by reading full VM data and used as a means of protecting against long incremental chains.

According to the best practice, in case of reversed incremental mode you should still run active full backup, though, on less frequent basis, say, 1 to 3 months.

I was trying to point out that with scheduled active full backup, it might occur that sometimes there are more than 1 .vbk (full) file in the backup repository. In this case, all of them, not only the latest one, will be copied to tape mediums by backup to tape job.

The tapes will be appended to as long as there is space, but will be overwritten when there is not enough space.

Correct.

Thanks.

[Dima P.]

The options are somewhat confusing if we just want to get the full backup to tape and offsite.

In this case - it’s up to you what backup method to use, as full backups are always picked up by tape job no matter what. I would suggest forward incremental if there is still a possibility you may need to enable incremental backup to tape as well.

So if I understand correctly, I need to set an retention on the pool. I have set "Do not protect data (cyclically overwrite tapes as required)." I think that will meet my needs. The tapes will be appended to as long as there is space, but will be overwritten when there is not enough space.

With this option VBR would recycle tape media when all tape storage would be in use, so it is not the best choice if your company has a strict retention policy to keep tapes for XX months, as tape backups could be overwritten by mistake (let’s say you forget to insert new clean media into library). I would rather advice you to use fixed media pool retention, like 12 months or so.

[echoe]

/---/

So if I understand correctly, I need to set an retention on the pool. I have set "Do not protect data (cyclically overwrite tapes as required)." I think that will meet my needs. The tapes will be appended to as long as there is space, but will be overwritten when there is not enough space.

Thank you. After reading this I realised that setting a manual retention would be helpful in my case. One of our client changes tapes manually in a standalone drive. Well, the GFS or whatever else is performed manually then and all I want from Veeam is to overwrite whatever tape there is inserted. So I thought "Do not protect data..." would be the most natural choice of retention for it. Wrong. What happened was that the next time the same tape was used, Veeam added to the previous media set and found quickly that there is not enough space on the tape. Now, after I chose "Protect data for 1 Days", only then I could get what I wanted. Let it be 1 day then, at least it overwrites the previous. (I also have set "Create new media set for every backup session" for Media Set properties.)

[veremin]

What happened was that the next time the same tape was used, Veeam added to the previous media set and found quickly that there is not enough space on the tape.

I've asked QA team to confirm this. For now the described behavior looks unexpected. Thanks.