High-level in that I’m not sure of the question to ask in order to get the answer I want…
I have an office where I’m given two Ethernet ports to what presents at 192.168.9.1 (on both ports) as an Ubiquiti UDM-Pro switch. It’s configured with d/l restriction of 30Mbit/s (actually 28) and apparently no upload restriction (it reports now on my Fritzbox router as 200.0 Mbit/s, but has reported as high as 780). At the moment I have the Fritzbox serving DHCP leases using one of the ports manually configured to 192.168.0.2. I’d like to make the best use of both ports with my server routing traffic between the local network and outside world via 192.168.9.1 and retiring the Fritzbox to provide wifi and voip to the office. The sever has 6 Ethernet ports available.
It seems I’m able to get higher throughput using both Ethernet ports together - using speedtest-cli to do a very crude test with two machines (192.168.9.2 & 192.168.9.3) hitting entre together, I get 40 Mbit/s d/l and similar relative improvement on the upload.
I’m given no IPv6.
Initially, I thought to bond a couple of the server’s ports and hope the switch would auto configure with 802.3ad - fat chance 
Then I had the idea of unbonding the ports and configuring each with it’s own IP (192.168.9.2 & 192.168.9.3) but even if I could somehow munge the two ports together, my research suggests this is a deadend because of the “weak end system model” or “weak host model” discussed here (which is well beyond me).
It does occur to me that I might ask for different subnets on each of the Ethernet ports I’m given, but I figure I’d need a good excuse for asking…
Any ideas of a way forward?
Many thanks.