V1 feedback

[nicolassimondiddi]

On the same boat here.

Opened a support ticket:
Veeam Support - Case # 07641525 - After upgrading VBR to latest version PROXMOX vm backup starts failing

Didn't get quite big help from support, just straight answer; 'this isnt supported' & no explanation (yet) why this worked in previous version :/

So we had to act on our own, as a workaround deployed Veeam Linux agent to continue backup of customers critical production server.

Downgrading to lower version of VBR (we thought about it also, but didnt choose that path).

Also from encrypted backups perspective we are now thinking about switching all VM's backup on ProxMox hypervisior with Linux agent backup approach.

How come Proxmox VM's backup don't support encrypted backups like Vmware&Hyper-V?

I wouldn't use veeam backup for proxmox for production use in the first place.
It's more a black magic poor implementation of a backup alternative for PVE than a real backup product for me, especially when you compare it to B&R for Hyper-V and Vmware.

I wouldn't use it for anything more than for a lab right now and non-critical data, however, PBS is not better.
PBE is a great product, it just misses the main feature for production use, a real backup solution (with better options than just full backup every time or that crap of PBS).

[Gostev]

Give a newborn some time before you start comparing it with a 15 year old, will ya

V1 capabilities of our Proxmox support already exceed what our VMware support offered in V4 for example, and it's adoption pace beats one of VMware V1 hands down, including some first very large customers using it in production, etc.

[nicolassimondiddi]

What I'm saying is that if we are honest, the proxmox plugin should still be a beta, but marketing is marketing and the criticism is on this point.

There are too many strange things for production use, not limited to:
- Why is a VM needed on host for backup? All other concurrent products don't need this
-- Why the vm storage should be on lvm-thin anyway? Doesn't make sense
- Why don't we have any reverse incremental or encryption option?
- Why don't we have the standard job scheduling capabilities?
- Why some file systems aren't supported? It doesn't make any sense as long as the host file system supports snapshoting

[Gostev]

Splitting this into a separate discussion since you have completely deviated from BTRFS support topic with all these new questions and comments

I've no idea why you think this should "still be in beta" and implying Veeam is somehow dishonest by not keeping it in beta, when objective facts like usage telemetry shows it to be a wildly successful offering widely used in production environments after just half a year since it went GA. So, many Veeam customers are finding a value in the current capabilities, even if you're clearly not one of them.

Answering your questions,

1/ Veeam targets large customers and hot add mode provides best backup and restore performance with Proxmox. Roughly half of our VMware customer bases uses hot add proxies for the same reason, deploying a proxy VM manually on their hosts; although VMware in particular provides even faster transport (Direct SAN access) for customers with shared storage.

2/ Reverse incremental backup mode is removed even from the main product in V13, based on its actual usage.

3/ Encryption, job scheduling: some of it are current/temporary limitation of the new framework we use to quickly add support for new hypervisors and this will be addressed going forward. Some other limitations were just about limiting the scope of V1, so we could ship it faster and gauge the interest.

4/ I believe there were technical reasons for file system support limitation, but here I need @PTide to comment.

[nicolassimondiddi]

Indeed, it has deviated
Still deviating, ouch, reverse incremental is really removed in V13?

How will this be handled? Because we use this for thousands of virtual machines...
The usage is low because most people just click on next without going further, this is not meaning there is no use case for reverse incremental.

We need the fastest restore possible at any time, and reverse incremental was the perfect way for this.

[PTide]

Hi,

-- Why the vm storage should be on lvm-thin anyway? Doesn't make sense

I don't believe there is such a requirement. Would you show us where did you see it, please?

- Why some file systems aren't supported? It doesn't make any sense as long as the host file system supports snapshoting

Unfortunately, BTRFS suppport status is technology preview. Doesn't make sense to support something what's not fully supported by the vendor. The product was never inteded to work with BTRFS and the limitation in the UG has been in place since the very first release.

Thanks!

[nicolassimondiddi]

Hi,
For the worker storage, see (where I originally found the solution) kvm-rhv-olvm-proxmox-ve-f62/no-storage- ... 95294.html

Indeed, BTRFS is in tech preview in Proxmox, but has existed on Debian for 10 years, it's old enough for us to try it

[PTide]

That's not true : )

[Gostev]

How will this be handled? Because we use this for thousands of virtual machines...
The usage is low because most people just click on next without going further, this is not meaning there is no use case for reverse incremental.

We need the fastest restore possible at any time, and reverse incremental was the perfect way for this.

Please use search as this was discussed in greatest details in the existing topics over the past months after it has been announced, including this common misconception that reverse incremental provides faster possible restore (it's not). You can search by poster as the explanation on the latter was from me personally.

[nicolassimondiddi]

The docs and the reality are sometimes two separate things

I can assure you that the worker creation assistant doesn't detect my local storage (which is "The default local storage")
I had to buy a usb ssd and create a lvm thin for the veeam worker.

[nicolassimondiddi]

Please use search as this was discussed in greatest details in the existing topics over the past months after it has been announced, including this common misconception that reverse incremental provides faster possible restore (it's not). You can search by poster as the explanation on the latter was from me personally.

I honestly don't read my Veeam forum digest as we receive XXX mails from Veeam every week as service provider / reseller.
Normally, I read the changelogs when planning the upgrades, not 1 year before

[PTide]

I can assure you that the worker creation assistant doesn't detect my local storage (which is "The default local storage")

If the storage is not BTRFS and is capable of creating snapshots, then it's worth opening a support case for investigation.

The docs and the reality are sometimes two separate things

SLA and ToS are built based on docs, hence we gotta stick with docs here : )

BTRFS may work for you and for Debian users, but we cannot oficially support it until Proxmox itself feels confident about the file system.

Thanks!

[nicolassimondiddi]

A support case is not worth it.
As I said, it's a lab, it's not what we use in production, it doesn't bring us money and we have some other things in Proxmox that block us from using it in production (beside the "official" support of BTRFS as you pointed it).

For now, we're gonna stick with Hyper-V/Veeam and we will look at this in some years (except if Microsoft does a broadcom like trickery with the licensing).

[PTide]

I understand. Let's take some notes together, what are the most crucial features for you?

Apparently you want these things:

- use VBR as a backup proxy/worker for PVE backup jobs
- standard job scheduling capabilities - what types of scheduling settings you think are missing?
- support for BTRFS

Anything else?

Thanks!