Question about best way to add new B&R Repository on host

[simcomit]

We have a Dell R510 server (I know it's getting old) running Windows Server 2016 as a Hyper-V host to primarily use for Veeam backups & replicas. The host OS is on a pair of 10k SAS disks in RAID1. It also has a set of ten 7200k NL-SAS disks in RAID10 on a PERC H710. The data array is (currently) formatted with NTFS. The host has a virtual Windows Server 2016 running on it with a full copy of Veeam B&R installed. I now need to figure out the best place to put the Repository for some server backups and have two ideas.

1) Create a new 2TB VHDX file on the RAID10 disks connected to the Veeam VM. I will format it with ReFS. I will then create a new repository and save the backups there. I may also make a second (or more) VHDX file for the replicas and additional backup repositories.

2) Create a new 2TB VHDX file on the RAID10 disks connected to the host server. I will format it with ReFS. I will then create a new repository and save the backups there. I may also make a second (or more) VHDX file for the replicas and additional backup repositories. Those VHDX files will be mounted on boot using a scheduled task.

The question is which would provide the best mix of speed and reliability? Will either method cause any unwanted problems (particularly the second idea)?

I would just go with the first option, but this host will possibly run other virtual servers (during a DR, or perhaps some light duty Linux systems like Observium) so I wanted to make sure the backups leave as much horsepower for other duties as possible, so I wondered if moving that disk use to the host outside of a VM process stack would be better or worse?

Ultimately I'm wondering what others are doing and what is the best practice, as well as if anyone has had trouble doing it one way or the other...

Thanks for the help!

[simcomit]

FYI, one of the reasons I wanted to put the Repositories on the host instead of connected to the local Veeam virtual server was in case that virtual server had any issues that wouldn't boot, I could still access the Repository from a Veeam B&R server at another location (or using the B&R Enterprise website, which we run in our Co-Location to manage all 4 sites and Veeam B&R servers I plan to deploy). The main reason for having a B&R server at each location is to keep jobs running if the network is out, but I'm trying to minimize the number of systems I have to license and maintain so I want to make sure I design it out as simply and flexible as possible.

[Regnor]

Why do you want to create the repository inside a VHDX? You'll have additional overhead and more complexity.
Just create the repository directly on the raid array of your Hyper-V server.

Btw. Don't put any production VMs on your backup server/Hyper-V. In case of a failure you'll lose both data and backups.

[simcomit]

I wanted to do it this way primarily because the Hyper-V host's data drive is already formatted NTFS and has over 6tb of data on it, and in order to use a ReFS formatted repository I'd have to move that data and running VMs off and back on the drive. I also wanted to keep the backup data in a VHDX container that way if I need to move to a new backup host I can just move that repository VHDX, as well as to keep the repository data on a different 'disk' than the other virtual VHDX files that host supports.

That said, now that I think about it (as I've slowly learned more about Veeam as I deploy it), is it even possible to just move a repository virtual disk (if I set it up the way I planned) and connect it to another Veeam repository server (probably another Hyper-V host)? Or is the only way to do this by using an existing fully running Veeam B&R server copy job to copy the data from the old repository to a new repository on another server? In other words, is it even possible to mount a VHDX file containing a repository (folder) from a broken Veeam "repository server" on a new "repository server" without having to use a copy job to re-seed the data?

As for using the backup server to run other VMs, as I mentioned I only plan to run non-critical workloads on there, as well as possibly a lab or recovery environment if I need to spin up a backed up server. I just wanted my backup infrastructure to be as host-agnostic as possible, hence why I installed Veeam on a virtual server and wanted to put the repositories in VHDX files so I could move it to another host if ever needed without having to do anything to the existing backup jobs and data.

[foggy]

In other words, is it even possible to mount a VHDX file containing a repository (folder) from a broken Veeam "repository server" on a new "repository server" without having to use a copy job to re-seed the data?

Yes, absolutely, it is just an ordinary VHDX with ordinary Windows volume, so you will be able to access the backups from any other Windows server.

[simcomit]

Thanks for the confirmation foggy! I really appreciate it, and now I know Veeam's more flexible than I feared. I'll have to look up the details on how I would mount an existing VHDX file on a new Veeam server and import/point it at the repository folder on the moved VHDX file...

Ignoring that, do you have any feedback on my original question about which of the 2 options would be best (practice, I know you can't really give a for-certain answer without knowing a lot more about the whole environment)?

[foggy]

I would prefer the second option, to eliminate the additional layer. We do not typically recommend storing backups inside guest OS.

[Poweruser]

and then think about adding:
ip, hostname, localhost or FQDN.
veeam behaves a bit strange there and allows to add all, but not linked.

[foggy]

It's just a matter of preference. What would linking give? Or what does not having it prevent you from?

[Poweruser]

if you add your host by multiple addresses (because an admin did not use the same naming scheme on every add) you willl have the same maschine multiple times included.
i fear this may create a strange behaviour if i add the same host multiple times.
will b&r check for identity of a host added multiple times? like localhost + vm1 + vm1.domain.com

[foggy]

No, it doesn't check and in fact this allows to workaround some network cases (where you need to use different interfaces of a single server, for example).

[Poweruser]

okay maybe it works as desired. nevertheless its a pity that i see my backupserver host next to my hyper-v host because the b&r runs all in one on one hyper-v host with repo, tape, hyper-v backup, b&r sql, ..
its a bit confusing but seems to be by design?

[foggy]

Do you mean the Managed Servers node under Backup Infrastructure? Unless you've added it twice, the host should be displayed under different subnodes there (once as Hyper-V host and once as backup server under Windows servers).

[Poweruser]

yes one is hyper-v and one windows server.
and if i install the agent server i think i will get a third one.
a bit confusing because i can access file system with server and hyper-v

[foggy]

I admit this might be a bit confusing, but this is expected, since a single server can have several roles.

[Poweruser]

hmm so if i buy and instlal the agent i will see my host 3 times?
running all veam b&r instances, hyper-v backup, agent backup, and host itself, and tape and repo proxy,..? :-D

maybe add a group per host(name), so i can see on per-host-basis whats all installed from veeam per host.
and so i see all the roles each hostname serves under one tree.