Backup Copy to Rotated USB

[netengineers]

Hello all,

I'm hoping to get a "best practice" recommendation on the best way to do backups to rotated media. Here is what I'm thinking:

1x "VMware Backup" Reverse Incremental job (Daily) that backs up up all VMs to our SAN
5x "Backup Copy" jobs (Mon/Tues/Wed/Thurs/Fri) that copy the data from the "VMware Backup" (SAN) to the USB media.

There would be 5x USB medias, each dedicated to their own respective day/repository. If I'm thinking correctly, this would allow us to not mess with the ForceCreateMissingVBK (DWORD) and ForceDeleteBackupFiles (DWORD) registry keys. Reason being that the Backup Job would only run on its dedicated day to its dedicated USB media, so the VBKs and other files from the previous backups would always be there.

Any red flags or things I've overlooked so far?

My real concern comes from the following:
1.) When creating a "Backup Copy" job the very first page talks about "Copy Every" and I'm not sure what to select. (I know I can block the backup copy from running from schedule section, but not sure how that plays into this)
2.) Will there be any "Transform" processes if I do not select the box "Keep the following restore points for archival purposes"? (Keeping in mind the original backup was done with Reverse Incremental)
3.) Will there be any "Transform" processes if I do select the box "Keep the following restore points for archival purposes"? (Keeping in mind the original backup was done with Reverse Incremental)

I'm pretty sure that a transform process on a USB drive would never complete in a reasonable amount of time, so I'm hoping to avoid that. Avoiding the registry keys above would give us better retention of data (doesn't have to be deleted every run) and could potentially increase backup performance (less data to copy) if transforms are avoidable.

So what are your thoughts?

[foggy]

1.) When creating a "Backup Copy" job the very first page talks about "Copy Every" and I'm not sure what to select. (I know I can block the backup copy from running from schedule section, but not sure how that plays into this)

The "Copy every" interval is responsible for which of the available restore points is being copied. For every copy interval, backup copy job copies only one restore point (the latest available). In your case you can set it to 7 days and each backup copy job will pick a restore point for each VM that is not older than 7 days and the subsequent backup copy job run will be triggered in 7 days.

2.) Will there be any "Transform" processes if I do not select the box "Keep the following restore points for archival purposes"? (Keeping in mind the original backup was done with Reverse Incremental)
3.) Will there be any "Transform" processes if I do select the box "Keep the following restore points for archival purposes"? (Keeping in mind the original backup was done with Reverse Incremental)

On reaching the specified number of restore points to keep, transform is performed during every job cycle, regardless of whether GFS retention is enabled or not. During this transform, the oldest increment is being injected into the VBK file. This does does not depend on the backup mode used by the source "normal" backup job.

[netengineers]

Thanks for that information, things are starting to make a little more sense. Lets see if I can get my remaining questions answered by using a more specific example.

Our daily Reverse Incremental backup runs Tuesday at 9:00PM.

• It is configured to retain 7 restore points

The Tuesday "Backup Copy" job is scheduled to only run during the 24 hour period that is Tuesday.

• It is configured to "Copy Every" 7 days.
• It is configured to retain 30 restore points
• No GFS retention

My Questions:
1.) Since the Reverse Incremental starts at 9:00PM and likely will not finish until Wednesday, will there be an issue with the "Copy Every" interval being set to 7 days? (Wednesday is technically the 8th day)
2.) If the Tuesday Reverse Incremental backup finishes on Wednesday, does that mean the Tuesday USB Media and associated "Backup Copy" job actually needs to be scheduled to run on Wednesday?
3.) At what point will the the transform operations on the "Backup Copy" side start to begin, at 7 days (Reverse Inc's retention) or at 30 days (Backup Copy's retention)?
4.) Does a "Backup Copy" simply do a copy/paste of the original reverse incremental .vbk/.vrb files until transforms are required?

• If not, how is it moving the backups to the destination? Is it Forward Inc or Reverse Inc "type" processing?

Maybe a little more background could be helpful. Currently we have 5 different Reverse Incremental Backup Jobs configured to backup directly to USB. Since naturally Reverse Inc requires more IOPs, the backups are taking way too long to complete and are running into the next day. This is because the USB drives can't handle the IOPs required for the VRB creation process. So I'm looking do the Reverse Inc to the SAN for faster performance and then get full restores points copied to the rotating USB drives.

Thanks again!

[foggy]

1. and 2. The Tuesday backup copy job will copy the latest restore point existing at the moment of its start (probably, the one created on Monday). Btw, along with the "Copy every" interval you can set the time when the interval should start on the particular day.
3. At 30 days backup copy job's retention.
4. Backup copy job is always incremental and it doesn’t copy backup files but rather synthetically creates required restore points in remote location from VM data in source backup repositories, copying only the required changed blocks.

While I totally understand the scenario, you should test whether USB drives are capable of handling the backup copy job load. Probably, a simple post-job task to copy backup files on the rotated media would be the way to go in your case.

[veremin]

In addition, if you don't want to rack your brains over appropriate backup copy job scheduling, you can use custom PS script that executes backup copy job, starting synchronization interval anew via "Sync Now" command. In the job settings, specify this script as a post job activity and select it to run on certain days only, using corresponding button.

Thanks.

[netengineers]

Just wanted to provide an update. I've started using the Backup Copy jobs and it seems like things are working as expected, though I'm wondering if that will continue to be the case. My concerns may very well not be valid and are just a result of my lack of knowledge on how Backup Copies work.

Can anyone tell me why the backup copies are in a constantly running state no matter how they are configured? (I had assumed they would only transition to a running state when scheduled/allowed to run).

Our Friday backup copy ran successfully, but now that the Friday USB media is disconnected it's associated Backup Copy job is just going back and forth (over and over) between a stopped (failed) and running state. Not sure if this is a cause for concern.

Any input?

[veremin]

Can anyone tell me why the backup copies are in a constantly running state no matter how they are configured? (I had assumed they would only transition to a running state when scheduled/allowed to run).

The backup copy job has a new "continuous" type - it always sits and wait for new restore point to come.

Our Friday backup copy ran successfully, but now that the Friday USB media is disconnected it's associated Backup Copy job is just going back and forth (over and over) between a stopped (failed) and running state. Not sure if this is a cause for concern.

If backup copy job runs successfully on desired day and fails on other days, you shouldn't be worry about getting the said warning/errors.

Thanks.