ZTM is an open source network infrastructure software for running a decentralized network. It is built upon HTTP/2 tunnels and can run on any sort of IP networks such as LANs, containerized networks and the Internet, etc.
ZTM lays the foundation for building decentralized applications by providing a set of core capabilities including:
- Network connectivity across Internet gateways and firewalls
- TLS-encrypted communication channels
- Certificate-based authentication and access control
- Decentralized application publishing and deployment
- Decentralized file discovery and data sharing
ZTM can be used in various settings ranging from a 2-node personal network connecting one's home and workplace to a 10,000-node enterprise network connecting offices and branches across the globe. Examples of applications that can leverage ZTM are:
- Remote access your home computer from anywhere in the world
- Share documents, pictures and videos within a group of people without the need of a big-tech social networking platform
- Private and secure P2P data transfer without the fear of eavesdropping
ZTM is written in PipyJS, a JavaScript dialect designed for Pipy (https://github.com/flomesh-io/pipy). Pipy is an open source programmable proxy software. Thanks to Pipy, ZTM has many unique features on top of the capabilities it offers:
-
Fast. HTTP/2 multiplexing is fast. And Pipy is fast. Like, C++ fast.
-
Secure. All traffic is encrypted by TLS and has identities via certificates. By using PipyJS, security policy can be easily customized to meet the requirements in your organization.
-
Highly customizable and programmable, since Pipy in itself is a general-purpose network scripting engine.
-
Portable. Choose your CPU architecture: x86, ARM, MIPS, RISC-V, LoongArch... Choose your operating system: Linux, Windows, macOS, FreeBSD, Android... ZTM runs anywhere.
The easiest way to get started is download the latest binary release of ZTM from our release page. If you prefer to have your own build from the source, you can follow the instructions in Build.
The official build releases of ZTM come in two forms of packaging: the CLI tool as a SEA (Single Executable Application), and the desktop application that wraps up the CLI tool and provides a GUI for desktop environments.
In this guide, we'll be only utilizing the CLI for setting up a simple mesh. For more guides, including the usage of the desktop app, please check out our Wiki.
A common setup consists of 3 nodes: 1 node running the Hub, the other 2 nodes running two Agents who wish to communicate with each other.
Data Center
+-------------------------------------------+
| Hub |
| (state in ~/.ztm/ztm-hub.db) |
+-------------------------------------------+
HTTPS | Port 8888 HTTPS | Port 8888
| |
------------|---------------------------------|--------------
| Firewall |
------------|---------------------------------|--------------
| |
| Internet |
| |
---------------------------- | ----------------------------
Firewall | Firewall
---------------------------- | ----------------------------
| | |
| | |
+--------------------------+ | +--------------------------+
| Agent @ Home | | | Agent @ Workplace |
| (state in ~/.ztm/ztm.db) | | | (state in ~/.ztm/ztm.db) |
+--------------------------+ | +--------------------------+
|
We'll only cover the setup of a Hub on Linux, since that's where they are usually run - a cloud-hosted Linux virtual machine.
Suppose you have a Linux box in the cloud, with a public IP address 1.2.3.4 and a public TCP port 8888. Start a Hub service by typing:
ztm start hub --listen 0.0.0.0:8888 --names 1.2.3.4:8888 --permit root.jsonYou might need
sudowhen executing the above command because it needs to install a service tosystemd.
Now the Hub should be up an running. Plus, a file named root.json should have been generated for us to allow endpoints to join our mesh.
Once the Hub gets up and running in the cloud, we can go on and add as many endpoints as we like to the mesh by using the generated permit file root.json.
An endpoint is just a computer running in various network environments with access to the Internet.
First, start an Agent on an endpoint computer that is going to join our mesh:
ztm start agentOn Windows, starting as a system service isn't supported yet. You'll have to do
ztm run agentinstead.
And then, join the mesh by saying:
ztm join MESH_NAME --as EP_NAME --permit root.jsonWhere MESH_NAME can be any name of your choice for identifying a mesh locally if you have many. EP_NAME is the name of your current endpoint seen by other endpoints in the same mesh. root.json is the permit file generated in our first step where a Hub is set up.
If everything works out, you can now check out the status of the mesh by typing:
ztm get meshOr look up for endpoints that already joined the mesh:
ztm get epFor detailed usage of the command-line tool, type:
ztm helpIf you prefer GUI, you can open your browser and point it to http://localhost:7777 right after command ztm start agent. You can join a mesh, find other endpoints, using apps and everything. Almost all functionalities ZTM provides are available from both the CLI and the GUI.
Repeat the above procedure for every endpoint in your mesh. Then, you will be able to manage your mesh via terminal or browser from any endpoint in the mesh.
Only connecting a bunch of endpoints as a mesh isn't very useful. What makes your mesh useful is the apps running in it. The official ZTM releases come with a number of builtin apps including:
- Tunnel - Establish secure TCP/UDP tunnels between endpoints
- Proxy - A SOCKS/HTTP forward proxy that takes in traffic from one endpoint and forward out via another endpoint
- Script - Execute PipyJS scripts remotely on an endpoint
- Terminal - Remote access to the shell on an endpoint
Third-party apps can also be installed. Also, new apps can be developed rather easily thanks to the PipyJS scripting capability of Pipy.
To get a list of all installed apps, type:
ztm get appYou can use an app from either the browser GUI or the command-line tool. On a terminal, one can access an app's CLI in a way like:
ztm APP_NAME ...To find out detailed information about using an app via CLI, type:
ztm APP_NAME helpHere's a recap of what CLI commands you need to do on each computer node.
Cloud-hosted VM
+---------------------------------------------------------+
| ztm start hub --names x.x.x.x:8888 --permit root.json | ---+
+---------------------------------------------------------+ |
| x.x.x.x:8888 | |
------------|--------------------------------|------------- |
| Firewall | |
------------|--------------------------------|------------- |
| | |
| Internet | | root.json
| | |
-------------------------- | -------------------------- |
Firewall | Firewall |
-------------------------- | -------------------------- |
| | | |
| | | |
+------------------------+ | +------------------------+ |
| ztm start agent | | | ztm start agent | |
| ztm join my-mesh \ | | | ztm join my-mesh \ | <--+
| --as home \ | | | --as workplace \ |
| --permit root.json | | | --permit root.json |
+------------------------+ | +------------------------+
PC @ Home | PC @ Workplace
For more information on the CLI, please refer to:
ztm help