Having at least one backup proxy server in each site when defining a replica job is recommended because the backup proxy servers are responsible for data processing and transfer. Having proxies in both sites enables a stable connection for VM data transfer across sites, as they handle the data compression, deduplication, and transfer processes. This setup ensures that data is efficiently moved from one site to another, thereby making replication more resilient and reliable. Proxies do not automatically restart replication after failure (A), remove the need for VM snapshots during transit (B and F), or enable automatic WAN acceleration (D), although they can work in conjunction with WAN accelerators if configured to do so. They also don't directly deduplicate data during transit across the WAN (C), although they do compress and optimize it for transfer.

no comment

C I Think The traffic from the source to the backup proxy is not optimized, for example, compressed or de-duped. Therefore, 100% of the backup data is transferred over this connection. Between the backup proxy and backup repository, optimized data typically about 50% of the source data size is transferred over this connection. https://helpcenter.veeam.com/docs/backup/vsphere/backup_proxy.html?ver=120