Hi I have couple of questions regarding the memory and disk usages in railway processes.

Railway is a nice place and has lower entry barrier, I like how things are working out here. However, after a while I noticed I am constantly trying to reduce memory usage on all my services to cut costs.

You are actually renting out your machines with good cpu and a substantial amount of memory + disk for containers (not DB). However, what you are charging for memory is a little bit vague for me.

If I rented an entry level GPU machine with 20 GB memory + 80 Disk and a good cpu it would cost 0.29$ on the average (L4 for instance). This would take 0.29x24x30=208.8$. This means I can get both GPU, Disk, Rapid Memory for a clean price. Though for 32cpu 32gb only I am payin around 100$ when I constantly occupy 5GB or so.

I know what you are doing needs to profitable but you are also charging the memory that I have no control of.

Below is my du output. And my dashboard shows around 3 GB usage and half of it is YOUR abstraction where I do not have control. If I scale I know this will be a small budget but considering the initiatives you have on hobby plan it makes less sense to switch to pro considering the usage based pricing of memory.

1.4G	/usr
146M	/root
15M	/var
2.5M	/app
2.1M	/etc
16K	/tmp
8.0K	/run
4.0K	/srv
4.0K	/opt
4.0K	/mnt
4.0K	/media
4.0K	/home
4.0K	/boot
0	/sys
0	/sbin
0	/proc
0	/lib64
0	/lib
0	/dev
0	/bin

I am willing to pay more for a good deal but these stuff wander in my head considering scaling up would cause me much comparing to other platforms. I don't want to churn but it would be good if you could revisit your memory pricing.

The least you can do is to partition the fast and slow memory for processes so that for stuff that I do not need rapid access I don't have to pay that much inside my container.