I think that some of the answers you’re looking for can be found in this post.
Zswap uses a portion of ram for swap and compression before going to the disk. Zswap seems to be the choice since the way zswap works is to move the oldest swap data to disk and keeps the newest and most used in ram. This is contrary to zram which never removes the oldest swap pages.
The link above provides this data.
Since version 3.11 the Linux kernel offers zswap support. If enabled,
the kernel will avoid swapping out pages to disk by compressing them
in ram. This will reduce disk I/O which can make systems with
mechanical disks feel more responsive or lower the wear of flash
based disks. The compressed memory pool is dynamically managed
by the kernel. Basically it creates an intermediary memory system
between the ram and the disk. This intermediate memory system is
more cpu intensive than any of the other two. Although the compressed
memory is slower than ram, it's still much faster than mechanical or
flash based disks. Latency, compression ratio and speed are dependent
on the compressor used with zswap.
Contrary to zram, zswap eventually swaps out the compressed pages to disk.
This behaviour can be configured by setting vm.swappiness inside
/etc/sysctl.conf. One of the advantages of zswap over zram is that zswap
requires no additional configuration in order to allow system hibernation.
Figured
I guess I’ll stick to Zswap
Thanks