This post was translated by Hermes Agent.

Three-line summary

1. A home server is not just a hobby for running services. It is personal infrastructure for operating my own data and automation.
2. When choosing the first machine, power usage, noise, stability, and RAM/SSD expandability mattered more than peak performance.
3. My recommended starting point is a low-power mini PC, 16GB of RAM, a 1TB NVMe SSD, wired LAN, Ubuntu Server, and a basic structure that can later expand with Docker.

Why I Built a Home Server and How I Prepared It

The reason I first wanted to build a home server was not grand.

My photos kept increasing and my iCloud storage was getting full. As a developer, I also thought it would be nice to have a mini PC that I could use like a small toy or lab machine.

Running tools like OpenClaw or Hermes continuously on my personal laptop felt awkward. If I keep services running on the laptop, I start worrying about battery cycles, and it also becomes unclear what to do whenever I take the laptop outside.

At first, the thought was simple.

It would be nice to have a small server at home that can stay on all the time.

But as I configured the mini PC and started building services one by one, the purpose became clearer.

I did not want to stop at simply storing files or running services. I wanted to record the flow of my daily life based on data from those services and view it through dashboards.

For example:

• how many memos I wrote in a day
• how much I ran
• how many movies or books I consumed
• which automation jobs succeeded or failed
• whether the services I use often are still alive

I also wanted to build small features and services quickly, expose them externally when needed, and restrict access when they should stay private.

What started as a simple mini PC gradually became a small piece of personal infrastructure for operating my own data.

In this post, I want to summarize why I built a home server, what criteria I used to choose the machine, and how I prepared the initial setup.

What I looked for when choosing the first home server machine

The first thing I had to decide was what kind of mini PC to buy.

A home server has different priorities from a normal desktop PC. It is less about short bursts of high performance and more about whether it can stay on reliably for a long time.

These were the criteria I looked at:

• low enough power usage to keep it running 24/7
• not too much fan noise or heat
• enough memory to run multiple services
• an SSD that can safely store the OS and service data
• easy recovery when something goes wrong
• smooth Linux installation and remote SSH management

At first, I thought CPU performance would matter the most. But after researching and running the server, I realized that power usage, noise, memory, storage, and recoverability mattered more.

A home server is not a machine that I use for an hour or two and then turn off. It stays on all day. So a configuration that is easy to keep running is more realistic than a slightly faster machine.

The specs of my home server

The mini PC I currently use is fairly low-end.

Starting with an overly powerful machine would have increased the cost, and it did not feel necessary for the services I wanted to run. I also thought that starting too grandly might make the whole thing less fun.

The specs are roughly as follows.

With this setup, I was able to run several personal Docker-based services such as Memos, Homepage, Uptime Kuma, and an HTML sharing service.

It is not a high-performance server. But it was enough as an entry-level personal home server, and the advantages of a low-power mini PC felt more important.

What I learned about CPU requirements

When I first started building the home server, I thought the CPU would be the most important part. After actually running it, I found that the CPU only needs to be “good enough.”

Typical web services such as Memos, Uptime Kuma, Homepage, and Filebrowser do not constantly use a lot of CPU. Most of the time they are close to idle, and the CPU is used only when a user connects or when a batch job runs.

For an entry-level home server, low-power CPUs such as Intel N100, N200, or N305 can be good choices.

That said, there are still things to check when looking at the CPU.

• Power usage: a home server runs all day, so high idle power usage becomes a long-term burden.
• Heat and noise: if the fan is loud or heat is high, it becomes annoying and you may end up turning it off.
• Instruction set and compatibility: with low-end or older CPUs, you need to check whether the software you want to run requires specific CPU instructions.

For example, older CPUs may not support AVX. Official MongoDB 5.0+ Docker images require AVX, so they may fail to run on those CPUs.

This kind of compatibility issue is easy to miss if you only look at the spec sheet. Before buying a mini PC, it is useful to search the CPU model together with keywords such as Linux, Docker, Proxmox, and MongoDB.

CPU model Linux
CPU model Docker
CPU model Proxmox
CPU model MongoDB AVX

A home server is a machine you buy once and keep running for a long time, so this kind of compatibility check can matter more than benchmark scores.

16GB as the minimum amount of memory

For a home server, memory matters more than CPU and is easier to feel in practice.

Each service may look light on its own, but memory usage grows quickly as Docker containers increase.

The types of services also become diverse:

• photo/media services
• memo services
• document or knowledge management tools
• file browsers
• monitoring tools
• batch and automation tools
• DB, Redis, search indexes
• AI agents or browser automation tasks

At first, I only planned to run a few services. But once you start operating a home server, you keep thinking, “Maybe I should run this too.”

My home server already uses around 8.3GB to 9GB of memory constantly.

For me, 16GB was a choice I did not regret.

You can start with 8GB, but once you add a few services, you will probably start thinking about memory. If you plan to run a home server for a while, I would treat 16GB as the default starting point.

SSD

Once you run services, data keeps accumulating.

At first, it may look like you only need space for the OS and Docker images. But over time, logs, DB data, uploaded files, backups, photos, and documents all keep growing.

In my case, storing photos and personal data was one of the reasons for building the home server, so I chose a slightly larger SSD.

For now, 1TB felt sufficient.

If you plan to store many large photos or videos, it is better to consider 2TB or more from the beginning, or prepare an external disk, NAS, or backup disk separately.

Other checklist items

When choosing a mini PC, CPU, RAM, and SSD are not the only things to check. You should also consider what might become inconvenient later.

RAM and SSD replaceability

Some mini PCs have soldered RAM or limited SSD expansion.

If possible, it is better to choose a model where the RAM can be expanded and the SSD can be replaced or added.

A home server may start small, but services and data tend to grow. A model with some expandability gives you more options later.

Wired LAN

For a home server, I think wired LAN should be the default.

Wi-Fi can work, but a server needs stability. If the network is unstable, service access, file transfers, SSH connections, and external access are all affected.

I would use at least 1GbE as the baseline, and 2.5GbE is also nice if the price is reasonable.

BIOS auto power recovery

It is useful if the machine can turn itself back on when power returns after an outage.

In BIOS, this feature is usually named something like:

1. Restore AC Power Loss
2. Auto Power On
3. Power On after Power Failure

A home server is often operated without a monitor or keyboard. If you have to press the power button manually after a power outage, it becomes surprisingly inconvenient.

So it is worth checking this option before buying or during the initial setup.

Linux installation reports

Even if the specs look good, operating the machine becomes painful if Linux has unstable networking or power management issues.

Before buying, I recommend searching the model name with keywords like:

model name Linux
model name Ubuntu Server
model name Proxmox
model name Docker homelab

Network cards, Wi-Fi, sleep mode, and boot issues are often mentioned in user reports.

A practical approach is to choose the mini PC brand or CPU line first, then compare how much the price changes depending on RAM and SSD capacity.

Things I prepared before installing the home server

Installing a home server requires more preparation than I first expected.

After installation, I can manage it through SSH. But during BIOS setup and OS installation, I often still need a keyboard and a screen.

This is especially true when the mini PC comes with Windows and I want to wipe it and install Linux.

After installation, the monitor and keyboard do not need to stay connected.

It is much more convenient to keep the home server in one place and manage it from a laptop through SSH.

Home server setup

OS

The mini PC I bought came with Windows installed. Since I planned to use it as a home server, I wiped it and installed Linux.

I did not need a GUI, and I did not need a Windows environment.

There are several OS choices for home servers:

• Ubuntu Server
• Debian
• Proxmox
• TrueNAS
• Unraid

Home server or NAS-focused operating systems such as Proxmox and Unraid are attractive, but they felt a bit too much for my immediate purpose.

Instead of starting with virtual machines, LXC, and storage design, I wanted to begin with a familiar Linux environment and add services one by one.

So I chose Ubuntu Server among Linux distributions.

The reasons were:

• Ubuntu feels stable and familiar
• it is widely used, so there is plenty of documentation
• it is easy to search for solutions when problems happen
• LTS versions have a long maintenance period
• unlike Ubuntu Desktop, it can run lightly without a GUI

Ubuntu Server is a server-oriented OS managed mainly through the terminal and SSH, rather than a desktop OS used with a graphical interface.

A GUI-less environment may feel awkward at first, but a home server is not a machine that you keep connected to a monitor. In that sense, this style actually fits well.

I will not cover how to create the OS installer USB or configure BIOS in detail here, because those steps vary by PC and there are already many resources online.

Basic setup after installation

I did not start running services immediately after installing Ubuntu Server.

First, I checked the basic settings needed to keep using it as a server.

1. Update packages

Right after installation, I updated the packages.

sudo apt update
sudo apt upgrade -y

A server is installed once and kept running for a long time, so it is good to start with up-to-date packages.

2. Install basic tools

I also installed tools that I use often while operating the server.

sudo apt install -y curl git vim htop ca-certificates

Roughly speaking:

• curl: checking HTTP requests or running install scripts
• git: managing configuration files or projects
• vim: editing simple files directly on the server
• htop: checking CPU and memory usage in the terminal
• ca-certificates: certificate bundle for trusting HTTPS sites

ca-certificates is something I rarely think about directly, but if it is missing or outdated, errors like certificate verify failed can happen.

It is needed when safely connecting to https:// URLs such as Docker install scripts, GitHub, or external APIs.

3. Set the timezone

I set the timezone to Korea time.

sudo timedatectl set-timezone Asia/Seoul

If the timezone is wrong, logs become confusing.

This becomes especially annoying later when cron jobs, batch jobs, or monitoring tools are involved. So it is better to set the timezone during the initial setup.

4. Reserve an IP address on the router

If the internal IP address of the server changes, annoying problems can happen.

For example, I might have to check the IP again every time I connect through SSH, internal service URLs may change, or external access settings may break later.

So I recommend setting a DHCP reservation on the router.

A DHCP reservation maps the server’s MAC address to a fixed internal IP, so the server receives the same IP even after rebooting.

You can check the server’s IP address with:

ip addr

Then, in the router admin page, find the server’s MAC address and assign the internal IP you want.

This makes future SSH connections and internal service access much easier.

SSH connection

Once the server installation and basic setup are done, the next step is preparing SSH access from a laptop.

A monitor and keyboard are needed during installation, but after that I do not want to keep touching the server directly.

I manage it from a laptop through SSH.

1. Connect by IP

The most basic method is connecting to the server’s internal IP.

ssh username@serverIP

For example, if the username is yeongsu and the server IP is 192.168.0.10, I can connect like this:

ssh [email protected]

At first, I can log in with the password I set during installation.

This method is simple, but it has drawbacks:

• it only works on the internal network
• I have to remember the IP
• I have to type the password every time

So I first verify that this works, then move on to SSH key-based access.

2. Connect with an SSH key

SSH key-based access uses a private key on the laptop and a public key registered on the server instead of typing the password every time.

The structure is simple.

My laptop: keeps the private key
Home server: registers the public key

The important rule is that the private key must never be shared, while the public key can be registered on the server.

If there is no SSH key on the laptop, it can be generated with:

ssh-keygen -t ed25519 -C "homeserver"

Using the default path usually creates these two files:

~/.ssh/id_ed25519
~/.ssh/id_ed25519.pub

id_ed25519 is the private key, and id_ed25519.pub is the public key.

The public key can be registered on the home server with ssh-copy-id.

ssh-copy-id -i ~/.ssh/id_ed25519.pub username@serverIP

This command adds the public key to this file on the server:

~/.ssh/authorized_keys

After that, I can connect with the SSH key without typing the server password every time.

If I set a passphrase on the private key, I may still need to enter that passphrase. That is not the server password; it is the password used to unlock my private key.

If SSH refuses to connect because of permissions, file permissions should be checked.

On the laptop:

chmod 700 ~/.ssh
chmod 600 ~/.ssh/id_ed25519
chmod 644 ~/.ssh/id_ed25519.pub
chmod 600 ~/.ssh/config

On the home server:

chmod 700 ~/.ssh
chmod 600 ~/.ssh/authorized_keys

SSH may reject key files if their permissions are too open.

3. Set an SSH alias

Typing this every time is annoying:

ssh [email protected]

So I can register an alias in the SSH config file on my laptop.

The file is:

~/.ssh/config

For example:

Host homeserver
    HostName 192.168.0.10
    User yeongsu
    IdentityFile ~/.ssh/id_ed25519

After this, I can connect with only:

ssh homeserver

At this point, the home server starts to feel less like a small PC with a monitor and more like a server managed remotely from my laptop.

4. SSH access from outside the home network

There is one important thing to note.

Connecting to an internal IP such as 192.168.x.x basically only works inside the same local network.

At home, this works:

ssh [email protected]

But from a company office or a cafe, this IP cannot be used directly. 192.168.x.x is a private IP address.

To access the home server from outside, a separate route is needed.

Common options are:

• port forwarding the SSH port on the router
• using a VPN such as Tailscale or WireGuard
• using Cloudflare Tunnel and Cloudflare Access

Later, I used Cloudflare Tunnel and Cloudflare Access so that I could connect to the home server externally without opening the SSH port directly on the router.

With that setup, I connect through a domain, pass Cloudflare Access authentication, and then reach the SSH service on the home server.

This belongs to external access and security, so I plan to cover it separately in a Cloudflare-focused post.

Closing

At this point, the most basic preparation for running a home server is done.

I prepared a mini PC, replaced Windows with Ubuntu Server, installed basic packages, set the timezone, reserved a stable internal IP on the router, and made it possible to connect remotely through SSH and SSH keys.

After this stage, the home server is no longer just a small PC that I touch with a monitor and keyboard. It is closer to a server that I manage remotely from my laptop.

Of course, no real service is running yet.

To operate services such as Memos, Uptime Kuma, Homepage, and Filebrowser, I still need Docker and Docker Compose.

External access also requires a separate structure involving domains, Cloudflare Tunnel, and Cloudflare Access.

At first, I simply thought I would buy a mini PC and run a few services. But while setting it up, I realized that a home server is also a small piece of infrastructure.

Choosing the machine, installing the OS, connecting remotely, running services, controlling external access, and monitoring status are all connected.

So in this first post, I focused only on the starting point: choosing the machine, installing the OS, doing the basic setup, and connecting through SSH.

In the next post, I will install Docker and Docker Compose on this server, organize service directories, and start running my first personal services one by one.