look. im getting real tired of people suggesting my contacts app doesn't need a full microservices setup. moved from $8 to $48 droplet because APPARENTLY running elasticsearch + mongodb + postgres + redis + kafka with full event sourcing uses "too much memory". anw what do they expect me to do? store phone numbers in a single table? like some kind of amateur??
my diagrams have 16 different services and yes they all need their own database. client keeps asking why saving a contact triggers 47 background jobs but they just don't understand best practices. keep getting OOM kills even with zero users but that's obviously because the health checks from my service mesh are generating too many events (each health ping needs to be stored in 3 databases for consistency). sure the app is currently crashing but that's only because we need more ram. might need to upgrade to the $96 instance to handle the write amplification from storing each crud operation as an immutable event
if anyone has solutions that don't involve removing my service mesh or suggesting sqlite i'm all ears. and no, "simplifying" isn't an option. i didn't spend 6 years studying distributed systems to build some legacy monolith
Just bought some pure, stunning Brazilian coffee straight from the terminal. What a time to be alive 😎
..Who wrote the captivating description?!
Also, Brazil mentioned !!! 🇧🇷
@ThePrimeagen@terminaldotshop@teej_dv