which you think could be the best practices for a proxmox server with several services in production [DISCUTION] - eviltoast

hi pals!

These days I am bothered by a question to improve the performance of my proxmox server. I hope you can advise me:

My Proxmos 8 24/7 is on an old dell optiplex 7010 4 x Intel® Core™ i3-3225 CPU @ 3.30GHz, 32gb RAM and it doesn’t give me problems normally it is at 20% CPU usage, 60% ram usage, 5% swap, and 12% disk usage and I have the following CT’s and VM’s configuration:

4 VMs (VM1 linux lite 5gb RAM) (VM2 win10 5gb RAM) (VM3 xpenology 4gb RAM) (VM4 OpenWrt 1gb RAM) all in production

7 LXC CT (CT1 512mg RAM, 1 docker service in production 1WP) (CT2 1gb RAM, 3 docker services in production POLR, CALIBER WEB, MATOMO) (CT3 2gb RAM, 8 docker services in production EXCALIDRAW, UPTIME-KUMA, CODE- SERVER, IPERF3, GUACAMOLE, HEIMDALL, LIBRESPEED, MEMOS) (CT4 2.5gb RAM 6 docker services in production 4WP, LITTLE-LINK, CALIBRE-WEB) (CT5 250mg RAM, 3 docker services in production 3VAULTWARDEN) (CT6 512mb RAM 2 services docker in production SPEEDTEST-TRACKER, OPEN-SPEED-TEST) (CT7 3gb RAM nextcloud in production)

My question is which of the following configurations could be best to increase the performance of my server?
1- create an LXC CT for each docker service, for example, 1LXC CT with RAM optimized for 1 each of my 5 WORDPRESS’s docker with the databases, 1LXC CT with RAM optimized for 1 guacamole’s docker , 1LXC CT with RAM optimized for matomo and so on for each service, or.

2- the same as I was doing until now, 1 LXC CT with optimized RAM and several docker services, or

3- 1 VM with good RAM and all the docker services inside?

I am also thinking about the fact that each LXC CT makes requests to the kernel, but also about the good use of cpu, ram and kernel resources.
Just in case I have a CGNAT so all my services go to the internet using cloudflare for CND domains and security, an external vps with wireguard and nginx.

thanks!