Setup your very own VPN in 10 minutes flat

Hey! Happy new year :-)

I've been looking to setup a personal VPN for a while, and the other week I discovered a rather brilliant project called PiVPN, which greatly simplifies the process of setting one up - and managing it thereafter.

It's been working rather well so far, so I thought I'd post about it so you can set one up for yourself too. But first though, we should look at the why. Why a VPN? What does it do?

Basically, a VPN let you punch a great big hole in the network that you're connected to and appear as if you're actually on a network elsewhere. The extent to which this is the case varies depending on the purpose, (for example a University or business might setup a VPN that allows members to access internal resources, but doesn't route all traffic through the VPN), but the general principle is the same.

It's best explained with a diagram. Imagine you're at a Café:

Everyone on the Café's WiFi can see the internet traffic you're sending out. If any of it is unencrypted, then they can additionally see the content of said traffic - e.g. emails you send, web pages you load, etc. Even if it's encrypted, statistical analysis can reveal which websites you're visiting and more.

If you don't trust a network that you're connected to, then by utilising a VPN you can create an encrypted tunnel to another location that you do trust:

Then, all that the other users of the Café's WiFi will see is an encrypted stream of packets - all heading for the same destination. All they'll know is roughly how much traffic you're sending and receiving, but not to where.

This is the primary reason that I'd like my own VPN. I trust the network I've got setup in my own house, so it stands to reason that I'd like to setup a VPN server there, and pretend that my devices when I'm out and about are still at home.

In theory, I should be able to access the resources on my home network too when I'm using such a VPN - which is an added bonus. Other reasons do exist for using a VPN, but I won't discuss them here.

In terms of VPN server software, I've done a fair amount of research into the different options available. My main criteria are as follows:

• Fairly easy to install
• Easy to understand what it's doing once installed (transparency)
• Easy to manage

The 2 main technologies I came across were OpenVPN and IPSec. Each has their own strengths & weaknesses. An IPSec VPN is, apparently, more efficient - especially since it executes on the client in kernel-space instead of user-space. It's a lighter protocol, too - leading to less overhead. It's also much more likely to be detected and blocked when travelling through strict firewalls, making me slightly unsure about it.

OpenVPN, on the other hand, executes entirely in user-space on both the client and the server - leading to a slightly greater overhead (especially with the mitigations for the recent Spectre & Meltdown hardware bugs). It does, however, use TLS (though over UDP by default). This characteristic makes it much more likely it'll slip through stricter firewalls. I'm unsure if that's a quality that I'm actually after or not.

Ultimately, it's the ease of management that points the way to my final choice. Looking into it, with both choices there's complex certificate management to be done whenever you want to add a new client to the VPN. For example, with StrongSwan (an open-source IPSec VPN program), you've got to generate a number of certificates with a chain of rather long commands - and the users themselves have passwords stored in plain text in a file!

While I've got no problem with reading and understanding such commands, I do have a problem with rememberability. If I want to add a new client, how easy is that to do? How long would I have to spend re-reading documentation to figure out how to do it?

Sure, I could write a program to manage the configuration files for me, but that would also require maintenance - and probably take much longer than I anticipate to write.

I forget where I found it, but it is for this reason that I ultimately decided to choose PiVPN. It's a set of scripts that sets up and manages one's an OpenVPN installation. To this end, it provides a single command - pivpn - that can be used to add, remove, and list clients and their statistics. With a concise help text, it makes it easy to figure out how to perform common tasks utilising existing terminal skills by conforming to established CLI interface norms.

If you want to install it yourself, then simply do this:

curl -L https://install.pivpn.io | bash

Of course, simply downloading and executing a random script from the Internet is never a good idea. Let's read it first:

curl -L https://install.pivpn.io | less

Once you're happy that it's not going to do anything malign to your system, proceed with the installation by executing the 1st command. It should guide you through a number of screens. Some important points I ran into:

• The static IP address it talks about is the IP address of your server on the local network. The installation asks about the public IP address in a later step. If you've already got a static IP setup on your server (and you probably have), then you don't need to worry about this.
• It asks you to install and enable unattended-upgrades. You should probably do this, but I ended up skipping this - as I've already got apticron setup and sending me regular emails - as I rather like to babysit the upgrade of packages on the main machines I manage. I might look into unattended-upgrades in the future if I acquire more servers than are comfortable to manage this way.
• Make sure you fully update your system before running the installation. I use this command: sudo apt update && sudo apt-get dist-upgrade && sudo apt-get autoclean && sudo apt-get autoremove
• Changing the port of the VPN isn't a bad idea, since PiVPN will automatically assemble .ovpn configuration files for you. I didn't end up doing this to start with, but I can always change it in the NAT rule I configured on my router later.
• Don't forget to allow OpenVPN through your firewall! For ufw users (like me), then it's something like sudo ufw allow <port_number>/udp.
• Don't forget to setup a NAT rule / port forwarding on your router if said server doesn't have a public IP address (if it's IPv4 it probably doesn't). If you're confused on this point, comment below and I'll blog about it. It's..... a complicated topic.

If you'd like a more in-depth guide to setting up PiVPN, then I can recommend this guide. It's a little bit dated (PiVPN now uses elliptical-curve cryptography by default), but still serves to illustrate the process pretty well.

If you're confused about some of the concepts I've presented here - leave a comment below! I'm happy to explain them in more detail. Who knows - I might end up writing another blog post on the subject....