What a remote access VPN does
A remote access VPN connects one device to a private network through an encrypted tunnel. This guide explains the parts, the protocols, and the weak points of the design. It also shows how zero trust network access changes the model.
The client calls the gateway
The VPN client on the laptop connects to the public address of the VPN gateway.
The gateway checks the user
The gateway asks an authentication server if the user name, password, and second factor are correct.
The tunnel carries packets
The client encrypts each packet for the office network and sends it through the tunnel.
The laptop is on the network
The laptop gets an office address. It can now reach each system that the office firewall does not block.
What a remote access VPN is
VPN means virtual private network. A remote access VPN lets one device, such as a laptop or a phone, join a private network from a different place. The device behaves as if it has a cable to the office.
The VPN does two jobs. It encrypts the traffic, so that the networks between the device and the office cannot read it. It also gives the device an address on the private network, so that internal systems accept its packets.
Remote access VPN and site-to-site VPN
A remote access VPN connects one device to one network. A person starts the connection and signs in.
A site-to-site VPN connects two networks, such as an office and a cloud VPC. Two routers or gateways keep the tunnel open at all times. No person signs in, and the devices on each side need no VPN software.
The two types use the same protocols. They are different in who connects, and in how the connection starts.
The parts of a remote access VPN
Each remote access VPN has three parts.
- The VPN client. Software on the device makes a virtual network interface. The operating system sends packets for the private network to that interface, and the client encrypts them.
- The VPN gateway. A server or an appliance at the edge of the private network accepts connections from clients. It decrypts the packets and forwards them to the internal network. People also call it a VPN server or a VPN concentrator.
- Authentication. The gateway must know who connects. It checks a password, a certificate, or a sign-in with an identity provider. Many gateways ask a directory or a RADIUS server.
Full tunnel and split tunnel
A full tunnel sends all traffic from the device through the VPN gateway, and that includes traffic for public websites. A split tunnel sends only the traffic for the private network. A split tunnel is faster for the user and puts less load on the gateway.
The main VPN protocols
The protocol defines how the client and the gateway agree on keys and how they encrypt packets. Three families are common.
IPsec with IKEv2
IPsec is the oldest of the three, and most operating systems and firewalls include it. The IKEv2 protocol negotiates the keys, and IPsec then encrypts the packets. IPsec has many options, so two products sometimes need careful configuration to work together.
TLS-based VPNs
A TLS-based VPN, also called an SSL VPN, uses the same encryption as HTTPS. It frequently runs on TCP port 443, so it passes through most firewalls. A tunnel over TCP is slower when the network loses packets.
WireGuard
WireGuard is the newest of the three. It runs over UDP and uses one fixed set of modern ciphers, so it has few options to configure. Each peer has a public key. WireGuard does not include user sign-in or a way to give out keys, so a product must add them.
The weak points of the design
The remote access VPN comes from a time when all company systems were in one building. The design has three weak points today.
A public listener
The VPN gateway must accept connections from any address on the internet. Attackers scan for these gateways and try each new vulnerability on them. A team must install each security patch quickly.
Access to the whole network
The VPN puts the device on the network. A user who needs one app can frequently reach hundreds of systems. If an attacker steals one VPN credential, the attacker gets the same reach.
One choke point
All remote traffic goes through the gateway. When systems are in several clouds and offices, traffic takes a long path through one site. If the gateway fails, all remote work stops.
How zero trust network access changes the model
Zero trust network access (ZTNA) keeps the encrypted tunnel and changes the rule for who can use it. A VPN asks one question at the door, and then it trusts the device. ZTNA asks a question for each app.
- Access for each app. A policy permits one group of users to reach one service. A user cannot reach a system that no policy names.
- Identity first. The user signs in with the company identity provider. When an administrator suspends the account, access stops.
- The device counts. A policy can require a managed device, an encrypted disk, or a specified country.
- No public listener. In many ZTNA designs, the gateway makes only outbound connections. An attacker finds no open port to scan.
- Many gateways. A small gateway sits in each network that has apps. Traffic goes directly to the correct network, not through one central site.
ZTNA is not automatically better. It needs a list of your apps and a decision about who uses each one. A team that skips this work gets a VPN with a new name.
How Firezone does remote access
Firezone is an open source zero trust network access platform built on WireGuard. You run the Gateways, and Firezone runs the control plane. Each part of the design answers one of the weak points above.
- Access for each Resource. A Resource is a DNS name, an IP address, or a CIDR range. A Policy gives one Group access to one Resource. All other access is refused.
- No inbound ports. Clients and Gateways make only outbound connections and find a direct path with NAT hole punching. You open no port on your firewall. See how hole punching works.
- WireGuard for each connection. The Client encrypts traffic end to end to the Gateway. Firezone adds the sign-in and the key distribution that WireGuard does not include. Firezone cannot decrypt your traffic.
- Split tunnel by default. Only traffic for Resources goes through Firezone. Public traffic uses the normal path of the device.
- More than one Gateway. Put two or more Gateways in a Site. If one fails, the Client moves to a healthy Gateway.
Run it beside the VPN you have
You do not need to replace a remote access VPN in one day. Firezone runs beside an existing VPN. Start with one CIDR Resource that mirrors the VPN, then add narrow Resources and move one app at a time.
An IPsec or TLS VPN, plain WireGuard, and Firezone
Plain WireGuard means the protocol with configuration files that you write and distribute. The table shows what each approach gives you for remote access.
| IPsec / TLS VPN | Plain WireGuard | Firezone | |
|---|---|---|---|
| User sign-in with an identity provider | PartialDepends on the product | NoKeys only | YesOIDC or an email code |
| Smallest unit of access | A subnet | The allowed IP ranges of a peer | One name, address, or port range |
| No inbound firewall ports | No | NoOne peer needs an open UDP port | Yes |
| Automatic key distribution | PartialCertificates or a sign-in | NoYou copy keys to each peer | YesNew keys at each sign-in |
| Failover between gateways | PartialWith added hardware or setup | No | YesTwo or more Gateways in a Site |
| Built into operating systems | YesIPsec in most systems | PartialIn the Linux kernel | NoInstall the Client |
| Works when UDP is blocked | YesA TLS VPN on TCP port 443 | No | No |
The table describes the usual design of each category. A specific product can be different. Read the documentation of the product before you decide.
Common questions
What is a remote access VPN used for?
A remote access VPN lets a person reach private systems from outside the office. Typical systems are file servers, internal web apps, databases, and remote desktops. The VPN encrypts the traffic on its path through the internet.
What is the difference between a remote access VPN and a site-to-site VPN?
A remote access VPN connects one device to a network, and a person signs in to start it. A site-to-site VPN connects two networks with a tunnel that stays open. The devices behind a site-to-site VPN need no VPN software.
Is a remote access VPN secure?
The encryption of a modern VPN is strong. The risk is in the design. The VPN gateway listens on the internet, and one stolen credential gives access to a full network. Multi-factor sign-in, fast patches, and narrow firewall rules decrease the risk.
Does a remote access VPN hide my browsing?
Only with a full tunnel. Then your traffic goes out through the company network, and the company can see it. With a split tunnel, traffic for public websites does not use the VPN.
Is WireGuard a remote access VPN?
WireGuard is a protocol, not a full product. It encrypts traffic between peers that know the public key of each other. A remote access product adds user sign-in, key distribution, and access rules to the protocol.