Question about deduplication

[verre]

Hi everybody, I've a question for you. Suppose I need to backup two vm in parallel using two proxies. The first proxy deduplicates the first machine and the second proxy makes the same with the second machine. Suppose that the two vm are quite similar... does someone deduplicate the two machine another time together?

[veremin]

Kindly, take a look at the sticky FAQ, should clarify your question. Thanks.

[verre]

Thank you Eremin. I had already read the FAQ. Indeed my doubt is: as I have understood the second deduplication between the two vm, take place on the "target side transport service", is it? So... I already know that the "transport service" could be installed on a Linux VM. Therefore a Transport Service on a linux vm can deduplicate and this is my last question: if a linux vm with transport service can deduplicate there is no differences between a linux vm with transport service and a windows proxy vm! I don't understand what a proxy windows has more than a transport service linux.

[foggy]

Not sure I understand your question. It is correct that target-side deduplication is performed by Veeam data mover agent (aka transport service) installed on the repository server. After being retrieved and processed by the source data mover agent installed on a proxy, data is being sent to the target data mover agent (repository), where deduplication between different VMs in a job is performed and data is then written to the storage.

[foggy]

If your concern about Linux proxies, then this thread might be helpful.

[verre]

Thank you foggy. My question is: the "Target Side Transport Service" deduplicates the data and it is installed in a Linux vm. Is it?
I don't understant what the proxy (windows) does more than a "Target Side Transport Service" Linux.
I understand that the "Source Side Transport Service" could be installed only on a Windows machine (we called that proxy) but no reason to install another proxy (another windows license) as a target

[foggy]

The proxy retrieves the data from the source datastore, dedupes/compresses it, and sends it over to the target storage. In terms of deduplication, it eliminates multiple copies of the same files stored inside VM, for example. Here's a bit more on that.

[verre]

Yes Foggy thank you. Could you check if this paint is correct?

https://drive.google.com/file/d/0B4V1PF ... sp=sharing

[dellock6]

Yes, the design is correct.
One note however: on Linux the transport executable is not running as a service (daemon in this case...), rather is dinamically deployed at each run and executed, and then removed when the job is over. That's why for example we require to enable ssh and have perl into the linux box, and the linux itself being reachable via ssh from VBR server.