How does the WAN Acceleration actually work?

[alphinantony]

Hi,
I'd like to know the actual working of WAN Accelerators. I went through all the guides and forums, but i'd really like to know which component is actually transferring the data from source backup repository to target backup repository when using WAN Acceleration used for a Backup Copy Job for example. Is it the Source WAN accelerator itself or Source backup repository or Gateway server, if the source backup repository is a CIFS share ? And what ports are involved throughout this session till the job is completed?

And again to which component its actually transferring the data to, is it Target WAN Accelerator, or Target backup repository, or Gateway server, if the Target backup repository is a CIFS share, and the ports involved?

It would be really helpful If someone could really explain the whole procedure? I know i'm asking a lot here.

Thanks!

[foggy]

Hi Alphin, in this case data is transferred directly between servers where WAN accelerators are installed. Have you reviewed this user guide section? All the involved ports are listed here. So the data path is as follows: source repository server (or gateway server) -> source WAN accelerator -> target WAN accelerator -> target repository server (or gateway server).

[alphinantony]

Thank you for the reply, foggy!
I've already gone through the user guide and the used ports section. If its like what you are saying then the traffic will flow between WAN Accelerators only, right? Only through the port 6165. No other ports are required for the a Backup Copy Job using WAN Accelerators. Suppose if there is a firewall in between these two WAN Accelerators, then from the start to the end of a Backup Copy Job only this port (6165) will be used (assuming the required services are installed in Target WAN Accelerator Server).



Then what is the use of these other ports(highlighted) mention in the screenshots? Are they used only for replication?

[foggy]

If we're talking about remote backup copy with WAN accelerators already deployed on both ends, then ports 6164 and 6165 are required. The highlighted ports are for communication between WAN accelerator and data mover installed on the repository server.

[alphinantony]

Okay! So two ports, 6164 & 6165.

And about the ports highlighted in screenshot, since they have given as FROM:WAN accelerator(source and target) TO :Backup repository(source and target) its a bit confusing. So i belive it should be read as :-

WAN accelerator(source) --> Backup repository(source)

WAN accelerator(target) --> Backup repository(target)

OR

WAN accelerator(source) --> Backup repository(source)
WAN accelerator(source) --> Backup repository(target)

WAN accelerator(target) --> Backup repository(target)
WAN accelerator(target) --> Backup repository(source)

I'm asking all this is because i recently deployed WAN acceleration with a firewall in between, and allowed only the 6165 and 6164. But got an error. I'll explain a part (Backup Copy job) in my scenario :-

Local (Source) Repository : CIFS share with a dedicated gateway server(IP : 10.250.15.60).
Source WAN Accelerator : It's installed on the gateway server mentioned above. So both roles(gateway & WAN Acceleration) are installed on the same server(IP : 10.250.15.60).
Offsite (Target ) Repository : CIFS share with a dedicated gateway server(IP : 10.251.5.209).
Target WAN Accelerator : It's the gateway server mentioned above. So both roles(gateway & WAN Acceleration) are installed on the same server(IP : 10.251.5.209).

When i run Backup Copy Job, the server(IP : 10.250.15.60) at the source tries to access the server (IP : 10.251.5.209) at the Offsite (target) on a port 2502 and fails(because its blocked in the f/w). This is visible from the Backup Copy Job statistics. And if i allow that port, the job continues and the WAN accelerators kick in and the job completes. This is the reason I'm confused. Does any other components (Source Repository or Gateway server) need to talk to their counter parts in the target site before transferring data via WAN accelerator?

[foggy]

Right, these ports are also used for direct communication between backup repository servers (gateways, in your case), since some data don't go through WAN accelerators (metadata, small VM files). Since these are the same servers where WAN accelerators are installed, you should open them for these servers as well.

[alphinantony]

Thank you foggy!
I think what you have mentioned above regrading what type of data goes through WAN Accelerators and what does not,is not there in the user guide, atleast not in the 'WAN Acceleration' section. There are a few reference about Backup Repositories involvement in a Backup Copy Job using WAN Accelerators in the 'Used Ports' section, but again they have mentioned it like the Backup Repositories (Source and Target) talks to each other only when WAN Acceleration is not used for a Backup Copy Job.



Another reference Backup Repositories is in the Backup Copy Architecture section in the User Guide for VMware vSphere, but its kind of vague and in the illustration just below it shows only the WAN Accelerator(Source) doing the data transfer. They could've atleast shown the defaults ports being used in all the diagrams in the User Guide.

[foggy]

Thanks, Alphin, we'll try to add some clarity in the description.