Health check on ReFS and protection from ransomware - macrium image guardian?

[folerx]

Hello
We now use Macrium site Manager for backup endpoints because presence of Image guardian technology which protect backup files from ransom attack.
But, It often happens that some of macrium incremental files have internal corruption even if option "verify" is checked. Or some files go into error state on disk.
Now we think about Veeam B&R as replacement. Here is health check for files in settings and email reporting. Nice.
1. Do we need this option if files is on ReFS file system which is "self-healing"?
2. Any technology like macrium image guardian for protecting backup files? Linux hardened repository is pain in a to configure. Image guardian is really good here.

[Dima P.]

Hello Daniel,

Thank you for your interest in our products!

1. Do we need this option if files is on ReFS file system which is "self-healing"?

I'd recommend to enable it all the time despite the underlying repository type. It checks the backup files being valid as well as the metadata being consistent, which can serve as a second layer of verification / notification.

2. Any technology like macrium image guardian for protecting backup files? Linux hardened repository is pain in a to configure. Image guardian is really good here.

Veeam Linux hardened repository or Object storage repository with immutability enabled is the best possible solution to address the backup immutability since it's controlled by a repository and not the backup application itself. Simply put, if something bad happens with your Veeam server, your backups will remain untouched as backup server cant do anything with the backups as long as those are immutable. Here is a good article: Protect against Ransomware with Immutable Backups: a Veeam Guide.

Let us know if you have any questions! Thank you!

[folerx]

If I dont use deduplication, any reason to use ntfs and not refs? fast clone on my mind to make less stress on hdd.

[grigatkis]

Veeam doesn't use deduplication, but rather block cloning.
https://helpcenter.veeam.com/docs/backu ... ml?ver=120

Which is based on ReFS, so you want to use that rather than NTFS if you can.

1. Do we need this option if files is on ReFS file system which is "self-healing"?

Definitely use health check - weekly if you can. I've had ReFS repositories go bad on me, which can be very problematic if it's not detected.

[Matts N]

Hello Daniel. I think some of us in here would argue that ReFS is many things, but not "self-healing". If you start getting problems with it, you often end up with having to reformat the entire partition. This is one reason Linux with XFS as file-system is getting more and more popular. And with the added possibility of immutable backups, which Windows lacks, it becomes the preferred choice. As for installation a hardened repository, there is an image in development to do most of the work for you: post529635.html#p529635

// Matts

[Gostev]

ReFS deployed on Storage Spaces, which is what it was originally designed to operate on, is indeed self-healing.

On simple volumes, Microsoft specifically recommends that redundancy is achieved by external means e.g. Backup Copy jobs in case of Veeam.

[grigatkis]

Oh, also regarding this point:

2. Any technology like macrium image guardian for protecting backup files? Linux hardened repository is pain in a to configure. Image guardian is really good here.

There's now an (experimental) ISO for easier deployment:
veeam-backup-replication-f2/managed-har ... 96192.html

Though I'll admit I haven't tried it out yet

[FedericoV]

ReFS and XFS are as good as the HW layer can be.
Since you are likely considering a parity RAID over high-capacity spindles, such as RAID 6 or 60, we know that performance is poor without using a Write-Back Cache. Now, the problem WBC is that in the event of power-loss, the FS will likely suffer from corruption. The workaround is to have a battery-powered HW that writes the cache to a flash chip. This way on the next reboot no data is lost and the FS is not corrupted.

Another element of reliability is to set disks to fail in "predicted failure" rather than waiting for them to die. In my lab I had a rack with a undersized power line causing multiple hard power-off every day. (Yes, it was a good unplanned test ). One of the servers in the rack had a FS corruption, and it was the one without the "predicted failure" active.
Also, in my experience, raid controllers are not made equally well, and there is no simple answer to that.

Once you have a good setup with the Linux Hardened repository, then the reliability and performance really pay off.
Now that ISO is available, I love this setup even more.

[Matts N]

ReFS deployed on Storage Spaces, which is what it was originally designed to operate on, is indeed self-healing.

On simple volumes, Microsoft specifically recommends that redundancy is achieved by external means e.g. Backup Copy jobs in case of Veeam.

Granted, we were using RAID-60 instead of Storage Spaces. The root cause to ReFS crashing was a bad backplane in the drive cage, which was a "hard" problem. I doubt Storage Spaces would have been able to recover from that, point being I would be careful calling something "self-healing" when there are so many factors that influence the outcome.

[Gostev]

Storage Spaces deliver redundancy across disks/servers/racks, so it's "self healing" in a sense that after a hardware failure you still retain access to all your data as if nothing happened, but it has reduced redundancy until an additional copy is recreated in place of the lost one to get that data back to the target/desired redundancy policy.

[Matts N]

In general I agree with you, Gostev. In our case I doubt Storage Spaces would have helped, since I suspect we had multiple corrupt drives due to backplane issues.