3 years worth of backups

[bhagen]

Going forward, we may be required to produce exactly what was displayed on our website ANY given day (or multiple days) for 3 year period.

Can somebody confirm that this stratagy will work:

Daily onsite incrementals of our webserver, with secondary backup copies being saved offsite
92 restore points (for a bit of wiggle room)
Backup to tape every 90 days (quarterly)

That will give us the ability to restore either an entire vm or specific files any day in the last 3 years, correct?

[veremin]

Actually, not, with the described strategy you will be able to restore:

- to any given day within the latest 92 days
- to any given quarter within the latest 3 year period (I assume retention will be set to 12 for quarterly backup)

To be able to restore to ANY given day within the last 3 year period, you will have to have daily backup job which retention is equal to 1095 or so - and currently UI allows you to go as far as 999 restore points only.

Thanks!

[bhagen]

Ugh...that's not good!

Is there a way to copy all 92 incrementals to tape every 3 months? I'm fine with having to go to the DR site and load a tape once a quarter, or in case we need to restore. But I really need an archive of every day.

Or, is there a better way?

[Mgamerz]

Couldn't you just have it write the incremental every day to tape?

[veremin]

Ah, I might have got your setup wrong, if the following scheme is true, then, everything will work fine:

Last 92 daily points stored in repository => Last 92 daily restore points copied to tape once a quarter

However, the mgamerz's suggestion looks simpler, indeed.

Thanks!

[bhagen]

"Last 92 daily points stored in repository => Last 92 daily restore points copied to tape once a quarter"

So would this be a "file copy to tape" or a "backup copy to tape" job?

[kewnev]

Correct me if I'm wrong but couldn't you do it this way..

Make sure you have 92 restore points set
Set the backup as incremental, with active full backups on Day 1 of months Jan, Apr, Jul, Oct.
Then do a file to tape job of the backup files in the repository on the last day of the month of Mar, Jun, Sep, Dec (although you need to do it before the full active backup starts the next day)

.. OR ..
Make sure you have 92 restore points set
Use Reverse incremental, and then you don't have to worry about timing the active full backups
Do a file to tape job of the backup files in the repository at 3 monthly intervals.

.. OR ..
A Backup Copy Job is probably even more convenient, because it uses forever-forward incremental. If you set the restore points to 92 days, then the oldest backup will merge with the full backup each day. Any time you write these files to tape will give you the last 92 days of backup.

Let me know if my thinkings are flawed..

[veremin]

So would this be a "file copy to tape" or a "backup copy to tape" job?

Both would work fine, the latter might preferable, though (easier VM backup management, simpler restore, etc.). Thanks!

[bhagen]

Thanks for the comments, everybody!

It seems my original plan will work. I had a bit of time this morning before another meeting, so I thought I'd document what I've done since it isn't really intuative, and maybe somebody else will benefit from the research in this thread.

I started daily incrementals of our webserver about 10 days ago, with "secondary target" jobs being sent to our offsite repository; both are set to 92 restore points
If I look in the VBR console, on the left side: Backups - Disk, then expand the backup job, I have 10 restore points
If I look in the VBR console, on the left side: Backups - Disk (Copy), then expand the "secondary target" job, I have 4 restore points. (That's because we're still establishing our DR site, and connectivity is still a bit spotty.)

Good so far.

Then as a test, I created a "backups to tape" job yesterday, pointing to the "secondary target" job.
The "backups to tape" job finished successfully, and I found this in the log: "Source backup files detected. VBK: 1, VBK map: 1, VIB: 3"

Then I tested a restore. It's a bit confusing figuring out how to restore, but I finally found what seems the best way to restore an entire VM from tape...which is what we'd need to do to satisfy the original requirement (to produce exactly what was displayed on our website ANY given day). The steps to create that job are:

In the VBR console, in the Home tab, click: Restore - vmware vsphere (not tape!) - restore from backup - entire vm restore - entire vm restore (again) - add vm - from backup - choose the tape backup job - click the name of the job you just added - click "Point..." on the right.

Once I finally get to this spot, I now see only the 3 VIB (incrementals). So I chose one of those, chose "restore directly from tape", and chose "restore to a new location". Then continued thru the process of telling it where to restore to.

That seems to work, and that satisfies our needs (as long as I can get those "secondary target" jobs working every day, but...that's a different issue!).

Again, thanks for the responses, and I hope documenting my journey can help somebody else!