I’d like to remind that zram has a cost when used and it’s not a magic bullet that generates usable memory out of nowhere.
It reserves some memory where it’s running, so it means your system has less memory to run, and it compresses unused memory on the fly when you run out of memory. This is setup as a swap device with a higher priority than a disk backed swap, so memory goes in there. The CPU cost is real and introduce lag that could be higher than swapping to a fast SSD.
I tend to o prefer zswap which is a bit different. It tries to compress data going to the swap before it happens, if it’s highly compressible, it keeps it in memory, otherwise it goes to the disk swap. It’s not necessarily reserving memory and will not hold data that can’t be compressed.
Those mechanisms can actually be good if you are really short on memory and you have a fast CPU compared to the storage backend throughput. The worse case IMO would be to use them while short on memory and use a quite old CPU, as they need memory to operate, you further reduce your own usable memory and increase your CPU usage a lot because of the memory pressure.