Watts are the real currency and ARM changes everything. I run a 2GB Ampere instance on Oracle Cloud free tier in Seoul. It uses 3 watts at idle. 128MB x86 is nostalgia, not efficiency.
128MB RAM still viable in 2024? My Alpine experiment
Following. I have a 256MB BuyVM slice in Luxembourg that runs WireGuard and nothing else. Cannot imagine a database in half that.
Guestbook is flat file now. SQLite would not start reliably under memory pressure anyway. MariaDB is gone. 71MB active. Status page updated.
This. I moved my blog to Hugo on a 64MB NAT VPS from Inception Hosting. No database, no PHP, no problems. The 128MB debate is already over if you accept static.
Have you tested with siege or wrk? I would like to see numbers before I believe 12 concurrent was the real peak. My own nginx on 1GB drops connections under wrk with keep-alive disabled.
I do not have wrk installed and I am not adding build dependencies to this instance. I used ab from another box: 847 req/s, mean 14ms, 0 failed on static files. Guestbook POST is 23 req/s. That is the number I have.
It's always DNS. Even when it's not. But here it is not DNS. It is that you are running edge Alpine which means you are rebuilding your world every update. I run stable and I have time for other hobbies.
Same here. I had a 128MB OpenVZ instance in 2012 that ran WordPress. It was miserable then and I do not miss it.
So the L in your LAMP stack was a lie from the start. This thread delivers.
I am arriving late. What happened to the status page? Is there a final config posted?