Is 1 GB RAM enough for self-hosting?

Yes, for lightweight services. A single Node.js app, a static site, or a small database fits comfortably in 1 GB. The moment you stack two or three services, you will feel the squeeze — and swap will become your lifeline.

How much RAM does a typical self-hosted stack use?

A reverse proxy (Nginx/Caddy) uses 20–50 MB, a Node.js app 100–400 MB, PostgreSQL 100–300 MB, Redis 30–100 MB, and Docker overhead 100–200 MB. A small full stack (app + DB + reverse proxy) typically settles around 600 MB to 1.2 GB.

Can I use swap instead of upgrading RAM?

Swap works as an emergency buffer, not a replacement for real RAM. It keeps your services alive during spikes, but relying on swap full-time makes everything noticeably slower because disk is 100x slower than RAM. Use it as a safety net, not a strategy.

When should I upgrade my VPS RAM?

Upgrade when you see these signs: OOM kills in dmesg, swap usage stays above 50% during normal operation, services respond slowly under moderate load, or you need to add another service and have no headroom. A good rule of thumb: if less than 20% of RAM is free during peak usage, it is time to grow.

Does Docker use more RAM than bare-metal installs?

Docker itself adds only about 30–50 MB of overhead. The real difference comes from running full container images versus minimal binaries. In practice, a well-configured Docker stack uses similar RAM to bare-metal, but gives you isolation and easy rollbacks in return.

What Hostinger VPS plan should I pick for self-hosting?

Start with KVM 1 (1 vCPU / 4 GB RAM) for one or two lightweight services. Move to KVM 2 (2 vCPU / 8 GB RAM) if you plan to run a full stack with Docker. The extra RAM headroom means fewer surprises as your setup grows.

This site contains affiliate links. If you purchase through them I may earn a commission at no extra cost to you. Full disclosure.

How Much RAM for Self-Hosting on a VPS

The short answer: 2 GB is the practical minimum for a self-hosted stack with more than one service. 4 GB gives you comfortable headroom for a database, an app, a reverse proxy, and Docker. Below 2 GB you will be juggling swap and crossing your fingers.

"RAM is the single bottleneck that decides whether your self-hosted setup feels snappy or sluggish. CPU matters for builds; RAM decides how many things you can run at once."

Real-world RAM usage by service

Marketing pages list system requirements in a vacuum. Here is what these services actually consume on a small VPS after they have been running for a few hours:

Service	Idle RAM	Under Load	Notes
Ubuntu / Debian base	80–150 MB	150–200 MB	Includes SSH, cron, syslog
Nginx (reverse proxy)	10–30 MB	30–80 MB	Scales with worker connections
Caddy	20–50 MB	50–100 MB	Slightly heavier than Nginx, auto-TLS included
Node.js app (small)	80–200 MB	200–500 MB	Depends heavily on app complexity
PostgreSQL	80–150 MB	200–400 MB	Shared_buffers defaults to ~128 MB
SQLite	5–20 MB	20–80 MB	Great for single-app setups
Redis	10–30 MB	30–100 MB	Scales with dataset size
Docker daemon	30–50 MB	50–100 MB	Overhead per container is small
OpenClaw	150–300 MB	300–600 MB	Node-based; depends on active agents

The math: stack your services

Add up the idle numbers for everything you plan to run simultaneously, then add 200–400 MB for OS overhead and buffers. That is your realistic minimum.

Example 1 — Minimal blog stack

OS base:          120 MB
Nginx:             25 MB
Node.js app:      150 MB
SQLite:            15 MB
─────────────────────────
Total idle:       310 MB
+ headroom:       200 MB
─────────────────────────
Comfortable at:   512 MB (tight) → 1 GB (safe)

Example 2 — Full self-hosted stack

OS base:          150 MB
Caddy:             40 MB
Node.js app:      250 MB
PostgreSQL:       150 MB
Redis:             30 MB
Docker overhead:   50 MB
OpenClaw:         250 MB
─────────────────────────
Total idle:       920 MB
+ headroom:       400 MB
─────────────────────────
Comfortable at: 1.3 GB → 2 GB (safe) → 4 GB (room to grow)

Example 3 — Docker-heavy multi-app

OS base:          150 MB
Traefik:           60 MB
App 1 (Node):     300 MB
App 2 (Python):   200 MB
PostgreSQL:       250 MB
Redis:             50 MB
Docker overhead:  100 MB
Monitoring:       100 MB
─────────────────────────
Total idle:     1,210 MB
+ headroom:       500 MB
─────────────────────────
Comfortable at: 1.7 GB → 4 GB (safe)

Swap: your emergency brake

Swap is disk space that the OS uses when RAM fills up. It prevents OOM kills but makes everything slower — reading from disk is roughly 100x slower than reading from RAM.

Always set up a swap file on a small VPS. It costs nothing and can save you from a midnight outage:

# Create a 2 GB swap file
sudo fallocate -l 2G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile

# Make it permanent
echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab

# Verify
free -h

Tune swappiness so the kernel prefers real RAM and only touches swap under pressure:

# Lower = prefer RAM, higher = more aggressive swapping
sudo sysctl vm.swappiness=10
echo 'vm.swappiness=10' | sudo tee -a /etc/sysctl.d/99-swappiness.conf

A swappiness of 10 means the kernel will keep using RAM until it is nearly full, then swap as a last resort. The default of 60 is too aggressive for VPS workloads.

How to check if you are running out of RAM

Do not guess — measure. These commands tell you exactly where you stand:

# Quick snapshot
free -h

# Live monitoring (press q to quit)
htop

# Check for OOM kills (these mean you ran out of RAM)
dmesg | grep -i "out of memory" | tail -5

# Per-process RAM usage, sorted by highest
ps aux --sort=-%mem | head -15

# Swap usage detail
swapon --show

Watch for these red flags:

Available memory below 200 MB during normal operation
Swap usage above 50% sustained over hours
OOM kills in dmesg — the kernel is killing your services to survive
Services restarting unexpectedly — often a silent OOM kill

When to upgrade vs. optimize

Before throwing money at a bigger plan, check whether your setup is wasting RAM:

Optimize first

Tune PostgreSQL — reduce shared_buffers and work_mem on small VPS instances
Limit Node.js heap — NODE_OPTIONS=--max-old-space-size=512 caps memory per process
Use Alpine images — Docker Alpine base images are 5 MB instead of 70+ MB
Stop unused services — systemctl list-units --type=service --state=running to audit
Drop GUI tools — phpMyAdmin, Portainer, and similar add 100–300 MB each

Upgrade when

You have optimized and still see OOM kills
You need to add a new service and have no headroom
Swap is your permanent resident, not an emergency visitor
You want to run Docker Compose stacks with 4+ containers

Starting fresh? The Hostinger KVM 1 plan (1 vCPU / 4 GB RAM) is enough for most beginner self-hosting setups. If you know you will run Docker with multiple services, jump straight to KVM 2 (2 vCPU / 8 GB RAM) — the price difference is small compared to the hassle of migrating later.

RAM cheat sheet by use case

Use Case	Minimum	Recommended
Static site or single lightweight app	512 MB	1 GB
WordPress or small CMS + DB	1 GB	2 GB
App + database + reverse proxy	1.5 GB	2–4 GB
Docker with 2–4 containers	2 GB	4 GB
Full self-hosted stack (5+ services)	4 GB	8 GB
CI/CD or build server	4 GB	8 GB+

The honest rule

Buy more RAM than you think you need — then use swap as your safety valve. RAM is cheap compared to the time you spend debugging OOM kills at midnight. A 4 GB VPS costs a few dollars more per month than a 2 GB one, and it saves you from the "just one more container" cliff that catches every self-hoster eventually.

If you are not sure where to start, read our guide on VPS sizing for self-hosting or check what to check before upgrading a VPS plan — sometimes the answer is optimize, not upgrade.

Is 1 GB RAM enough for self-hosting?: Yes, for lightweight services. A single Node.js app, a static site, or a small database fits comfortably in 1 GB. The moment you stack two or three services, you will feel the squeeze — and swap will become your lifeline.
How much RAM does a typical self-hosted stack use?: A reverse proxy (Nginx/Caddy) uses 20–50 MB, a Node.js app 100–400 MB, PostgreSQL 100–300 MB, Redis 30–100 MB, and Docker overhead 100–200 MB. A small full stack (app + DB + reverse proxy) typically settles around 600 MB to 1.2 GB.
Can I use swap instead of upgrading RAM?: Swap works as an emergency buffer, not a replacement for real RAM. It keeps your services alive during spikes, but relying on swap full-time makes everything noticeably slower because disk is 100x slower than RAM. Use it as a safety net, not a strategy.
When should I upgrade my VPS RAM?: Upgrade when you see these signs: OOM kills in dmesg, swap usage stays above 50% during normal operation, services respond slowly under moderate load, or you need to add another service and have no headroom. A good rule of thumb: if less than 20% of RAM is free during peak usage, it is time to grow.
Does Docker use more RAM than bare-metal installs?: Docker itself adds only about 30–50 MB of overhead. The real difference comes from running full container images versus minimal binaries. In practice, a well-configured Docker stack uses similar RAM to bare-metal, but gives you isolation and easy rollbacks in return.
What Hostinger VPS plan should I pick for self-hosting?: Start with KVM 1 (1 vCPU / 4 GB RAM) for one or two lightweight services. Move to KVM 2 (2 vCPU / 8 GB RAM) if you plan to run a full stack with Docker. The extra RAM headroom means fewer surprises as your setup grows.