unclear vm proxy behaviour

[real_tarantoga]

Hello,
until now we were backing up our VMs (vSphere 7) using VAS v10/v11, SMB shares on Windows 2019 Repositories (ReFS formatted) and Proxy VMs
The performance was not brilliant, but all jobs finished in time ...
The jobs even sayed they were using "fastclone" with "hotadd" ?!? I thought fast cloning is only available with all LUNs presented to the proxy server?

After a session with our supporter/reseller, they adviced me to switch the repository mode to "Direct Attached Storage" instead of using CIFS.
Because only with DAS mode we would profit from ReFS advantages ...
So I did and the performance is gone down. SMB is faster than the DAS mode on the same repository server.

They also declared, it would be better to use physical proxies - so I installed 2 HP blade servers with Windows 2019; the storage guy made all LUNs configured to ESXi hosts visible to these new physical proxies - and they were never used for backup!
I disabled all but these 2 physical proxies and the jobs failed or switched over to NBDtransport mode.

Before I will open a support case, I would check up with You if I'm just to stupid, missing essential technical knowledge to these issues:
- physical proxy with SAN-presented VM LUNs never uses fast cloning - why?
- why SMB can be faster than the recommended DAS on ReFS transfer mode?
- why are my proxy VMs saying, that tthey use "fastclone" ?

I really suppose that I'm not understanding the correct configuration ... but I read the knowledge base articles and the VBP guide (https://bp.veeam.com/vbr)

Thank You!
Stefan

[HannesK]

Hello,
sure, fastclone works with hotadd. fastclone is a repository functionality (well, with SMB it's the gateway server running on the proxy, but that's the technical background).

Fastclone with REFS works over SMB 3. And yes, support is correct by recommending DAS over SMB (CIFS doesn't work with fastclone. The protocol is dead since a decade).

For direct SAN backup: the LUNs must be visible in storage management as offline. If that's not the case, then something was misconfigured. https://www.veeam.com/kb1446

- physical proxy with SAN-presented VM LUNs never uses fast cloning - why?

the proxy is irrelevant. The repository is doing fastclone: https://helpcenter.veeam.com/docs/backu ... ml?ver=110

- why SMB can be faster than the recommended DAS on ReFS transfer mode?

corner case. hard to say without repository information. I have never seen that.

- why are my proxy VMs saying, that tthey use "fastclone" ?

probably it's not the proxy. hard to say as the repository information is missing. There could be something wrong with that server (e.g. tasks settings or something like that)

By the way: which storage (vendor / model) are you using? Backup from storage snapshots might be an option for you. https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Best regards,
Hannes

[real_tarantoga]

Thank You, Hannes!
Initially SMB shares were configured on the Windows 2019 Repository Servers, ReFS formatted DAS at these servers. Now it is changed by re-adding the same repository as a new backup repository with the option "direct attached storage"
So, that is the correct way?

The LUNs for VMware are all presented via SVC. The hospital will never pay the storage snapshot licensing costs, unfortunately.
The repository LUNs are NetApp or HDS via SVC ... no storage snapshot option. Always a question of "who will pay for it", actually nobody.

Just in this minute the storage guy called me to say, there's an issues to clear with IBM ... so perhaps I'm not as stupid as that.

I will give more feedback later.
Thank You again for now!
Stefan

[real_tarantoga]

Good morning, Hannes!
The LUNs are now correctly mapped and it works.
After reading your proposed link I will also recheck the repository configuration.

Regards!
Stefan