Coolify vs Dokploy vs CapRover: Self-Hosted PaaS on a VPS (2026)
I ran Coolify, Dokploy and CapRover side by side on a 4GB Hetzner VPS. Here is the honest breakdown of idle RAM, Docker Compose support, pricing and which one I now reach for first.
Side-by-side cloud platform comparisons
I ran Coolify, Dokploy and CapRover side by side on a 4GB Hetzner VPS. Here is the honest breakdown of idle RAM, Docker Compose support, pricing and which one I now reach for first.
The three headless CMSes everyone shortlists — Strapi, Directus, Payload — compared on what decides your VPS bill: RAM footprint, real monthly cost, extensibility, and migration risk. With hands-on picks by scenario.
Self-hosted error tracking no longer means donating half your VPS to Sentry. A hands-on 2026 comparison of Bugsink, GlitchTip, and self-hosted Sentry by RAM footprint, cost, install pain, and which one belongs on your $5 box.
A hands-on comparison of Woodpecker CI, Drone, and Concourse for self-hosted CI/CD on a VPS in 2026 — RAM footprint, licensing, learning curve, and which one to install first.
I installed k3s, k0s, MicroK8s, and Talos on identical Hetzner CX22 VPS boxes for three weeks. Real memory numbers, install times, the gotchas nobody mentions, and which one I picked for client work in 2026.
I installed Plane, OpenProject, and Leantime on the same Hostinger VPS, gave each one a week of real work from our 7-blog ops, and tracked RAM, response time, and friction. Here is what actually held up — and what surprised me.
Hands-on comparison of four self-hosted log aggregators on production VPS workloads. Real RAM measurements, decision framework, and the platform I picked for our seven aggregator sites.
I spent 3 months testing Documenso, Docuseal, and OpenSign side-by-side on the same VPS. Here are the resource numbers, deployment times, and the one I run in production for Wardigi today.
After running all three on a Hetzner VPS for 90 days with 18,400 documents, here's which self-hosted DMS actually earns its keep in 2026 — with real OCR benchmarks, RAM footprints, and verdict by use case.
Six weeks running all three self-hosted ChatGPT alternatives on Hetzner VPS — RAM benchmarks, RAG accuracy on real documents, and which one my team kept.
A hands-on 2026 comparison of three self-hosted LLM inference engines — when to pick Ollama, vLLM, or LocalAI for your GPU VPS, with throughput numbers, hardware sizing, and cost math.
Real benchmarks and operational tradeoffs from running Qdrant, Weaviate, and Milvus on small VPS plans. Latency, memory footprint, cost, and which one to pick for your RAG workload in 2026.