VPN Terminology Guide: Subscriptions, Nodes, Protocols & Split Tunneling
This guide explains subscriptions, nodes, route types, protocols, and split tunneling, plus the difference between global and rule-based modes, so beginners can make sense of their VPN client.
Subscription link: imports connection settings
In a VPN client, a subscription retrieves connection settings, while nodes are the connections you can choose from those settings. The protocol defines how the client and server communicate, and split tunneling determines which requests use a node. A subscription on a bill, by contrast, refers to a billing plan. Updating a client subscription refreshes the settings; it does not renew a plan or turn on the connection.
- Copy the subscription link for your current client from the service dashboard. In the client's Subscription or Configuration section, choose the option to import from a URL. Don't paste it into a server address field for an individual node.
- After importing, open the configuration and make sure the node list has loaded. If parsing fails, check the import type and whether the client supports that configuration format. If the request times out, check whether your current network can reach the subscription URL.
- Select a configuration and node, then connect. If the client also asks you to enable the system proxy or grant system VPN access, follow the prompts for your platform. Then test with the app you intend to use; a successful import does not mean the app is routing through the proxy.
Nodes and route types: the exit isn't the whole path
The region in a node name usually identifies its exit location. For websites that determine location from an IP address, what matters is the exit IP actually used for the request—not the name or flag. The route is how your device reaches that exit. Nodes with the same Japan exit can still differ in their entry point, transit networks, and congestion at different times.
| Name | What it means | What it doesn't mean |
|---|---|---|
| Direct node | The client connects directly to the proxy server without first going through a relay entry point provided by the service. | It doesn't mean there are no carrier routers along the way, and it isn't the same as “direct” in split-tunneling rules. |
| Relay node | The connection reaches an entry point first, then the service forwards it to another node or exit. | A nearby entry point doesn't guarantee a smooth route beyond it. |
| IEPL dedicated line | An international Ethernet private line. In a relay setup, it may carry traffic over part of the route between the entry and exit points. | A node label alone can't tell you which part of the route uses a dedicated line, let alone guarantee one from your device all the way to the website. |
“Direct” in a rule means traffic that matches it leaves without using a proxy. That's different from connecting to a direct node. To compare routes, test the same app at your usual times and check loading speed, sustained performance, and disconnections. Client latency checks depend on the test target and method, so they don't replace real-world testing or measure bandwidth.
Proxy protocols: the client and node must match
A protocol is a set of rules for establishing a connection and exchanging data. It isn't an exit region or a route quality rating. Even if your client supports a protocol, you can't simply change the protocol name on an existing node and expect it to work: the server protocol, authentication details, and transport settings all need to match.
- Shadowsocks: An encrypted proxy protocol. If importing fails, check not only that your client supports it, but also that the encryption method and credentials match.
- VMess and VLESS: These are different protocols, and their configurations aren't interchangeable. VLESS doesn't provide transport encryption by itself, so check the actual configuration to see whether protections such as TLS are in use.
- Trojan: Typically establishes connections over TLS. Enter the server address, the name used for certificate verification, and authentication details as specified by the server configuration.
- Hysteria2 and TUIC: Both are based on QUIC and typically require UDP connectivity, but they are different protocols. If a configuration connects at home but times out on a particular public network, compare it with another transport configuration from the service provider to check whether that network restricts UDP.
Whether desktop and mobile clients can use the same subscription depends on the formats and versions they support, not just the operating system. Being able to import a configuration doesn't mean every node in it will work. If only some nodes fail, first check whether your current client version supports their protocol and transport combinations.
Global mode vs. rule-based split tunneling
Global mode typically routes traffic captured by the client through the currently selected proxy or policy. Rule-based mode uses conditions such as domains and IP addresses to decide whether traffic goes through the proxy, connects directly, or is blocked. Many clients apply the first matching rule in order, but check the client's documentation; if nothing matches, the default policy at the end also matters.
“Global” doesn't mean every request from every app on your device goes through the proxy. A desktop system proxy generally applies only to apps that follow that setting; some apps may use their own network stack. With virtual network adapter routing, coverage also depends on routes and exclusions. On mobile, even when the system indicates that a VPN is active, check whether the client excludes certain apps. Conversely, direct connections in rule-based mode may be exactly what the configuration is designed to allow.
If the homepage loads but sign-in or images fail, don't check only the main domain shown in the address bar. The page may also request resources from separate sign-in services or static asset domains. If the client provides connection logs, check where those requests went while reproducing the issue.
Checking for DNS leaks and connection issues
DNS resolves domain names to IP addresses. DNS queries and actual web connections don't necessarily take the same route: a webpage may use a proxy while DNS is handled by the local network or a resolver selected by the browser. Investigate a possible leak only when queries that should follow the proxy policy are sent to an unexpected resolver. A local resolver deliberately configured for direct connections shouldn't be considered a leak just because it doesn't use the proxy.
- ✅ Test the connection in the browser or app you actually plan to use. An IP-check website only shows the exit used by that website's request; it can't confirm that other apps or DNS queries use the same route.
- ✅ If your client provides connection logs, check the matched rule and outbound policy for the target domain. If name resolution fails, also check the client's DNS settings and whether the browser has its own Secure DNS enabled.
- ✅ Check the resolver and DNS outbound route against the client documentation. A test page showing recursive resolvers and a web exit in different regions isn't, by itself, proof of a leak; what matters is whether DNS followed the intended route.
- ❌ Don't change the node, routing mode, and DNS settings all at once. Change one setting at a time, then repeat the same action in the same app so you can tell which change affected the result.
For your first connection, start with the service provider's default configuration. Once it works, adjust routing or DNS settings to suit your needs. Still need to install a client? Start with the getting started guide. Comparing node names and regions? See the routes page.