In addition to those, you can get some improvement (more so on a full node than a low-mem witness node) by using zswap (comressed memory). I don't have a specific writeup for that though.
If you are trying to use a low-mem witness node in less than 8 GB or a full node in less than 16 GB it is probably hopeless for now though. Upcoming improvements should make that possible again.
https://steemit.com/witness-category/@abit/best-practice-running-steemd-v0-16-0
https://steemit.com/witness-category/@bhuz/steemd-v0-16-0-and-chainbase-i-o-issues-and-possible-solutions
In addition to those, you can get some improvement (more so on a full node than a low-mem witness node) by using zswap (comressed memory). I don't have a specific writeup for that though.
If you are trying to use a low-mem witness node in less than 8 GB or a full node in less than 16 GB it is probably hopeless for now though. Upcoming improvements should make that possible again.
Thanks for the info.
A bit more info about zswap can be found here: https://steemit.com/witness-category/@aizensou/witness-update