So my goal is to reduce plot size as much as it is possible (so it takes less traffic to transfer via network). The conditions are:
It can be restored after moving
(Restoration_time + network_trans_time) < (New_Plot_genTime)
&&
(Restoration_time + network_trans_time) < (Uncompressed_plot_network_trans_time)
Is there any know ways to do so?
(yes, I tried myself and got into the hairy math behind the PoS, but managed to reach only 5-8%)
Because the plot files are essentially random data they are not really compressible. If you are looking to speed up transport on your internal network I would recommend looking at adopting 10 GB Ethernet. If you’re looking at transport over the Internet, sorry to say it’s never going to be great.
I tried compressing one once, it got bigger. FWIW, we have 1GB and 10GB internet speeds, so if you can get that it might work for you to someone who has similar, however 10GB ethernet looks like a go for local xfer. You might get 1000MB/s or so I read from tests.
Well… Anyway, thank you for your answers. 
Though, if I ever find a way to reach some comprehensible numbers of reduce, I’ll post it here.
If I remember correctly the Chia paper mentioned table1 could be potentially omitted and traded for processing for a 13% space saving.
If you’re doing all this manually, you can zip in many parts and then once transferred, unzip them.