Domain, SSL certificates, and load balancer
In order to have a secure LiveKit deployment, you will need a domain as well as a SSL certificate for that domain. This domain will be used as the primary endpoint for LiveKit clients, for example: wss://livekit.yourhost.com
. The SSL certificate must be signed by a trusted certificate authority; self-signed certs do not work here.
You will also need to set up HTTPS/SSL termination with a load balancer or reverse proxy.
If you are using TURN, then a separate TURN domain and SSL cert will be needed, as well.
Improving connectivity with TURN
Certain corporate firewalls block not only UDP traffic, but non-secure TCP traffic, as well. In those cases, it's helpful to use a TURN server. Here's a good resource if you're interested in reading more about how TURN is used.
The good news is LiveKit includes an embedded TURN server. It's a secure TURN implementation that has integrated authentication with the rest of LiveKit. The authentication layer ensures that only clients that have already established a signal connection could connect to our TURN server.
TURN/TLS
To firewalls, TLS traffic looks no different from regular HTTPS traffic to websites. Enabling TURN/TLS gives you the broadest coverage in client connectivity, including those behind corporate firewalls. TURN/TLS can be enabled with:
turn:enabled: truetls_port: 5349domain: turn.myhost.comcert_file: /path/to/turn.crtkey_file: /path/to/turn.key
LiveKit will perform TLS termination, so you will have to specify the certificates in the config. When running multiple LiveKit instances, you can place a layer 4 load balancer in front of the TCP port.
If you are not using a load balancer, turn.tls_port
needs to be set to 443, as that will be the port that's advertised to clients.
TURN/UDP
As QUIC (HTTP/3) gains adoption, some firewalls started allowing UDP traffic to pass through port 443. In those cases, it helps to use TURN/UDP on port 443. UDP is preferred over TCP for WebRTC traffic, as it has better control over congestion and latency. TURN/UDP can be enabled with:
turn:enabled: trueudp_port: 443
Configuration
For production deploys, we recommend using a config file. The config file can be passed in via --config
flag, or the body of the YAML can be set with a LIVEKIT_CONFIG
environment variable.
Below is a recommended config for a production deploy. To view other customization options, see config-sample.yaml
port: 7880log_level: infortc:tcp_port: 7881port_range_start: 50000port_range_end: 60000# use_external_ip should be set to true for most cloud environments where# the host has a public IP address, but is not exposed to the process.# LiveKit will attempt to use STUN to discover the true IP, and advertise# that IP with its clientsuse_external_ip: trueredis:# redis is recommended for production deploysaddress: my-redis-server.name:6379keys:# key-value pairs# your_api_key: <your_api_secret># When enabled, LiveKit will expose prometheus metrics on :6789/metrics#prometheus_port: 6789turn:enabled: true# domain must match tls certificatedomain: <turn.myhost.com># defaults to 3478. If not using a load balancer, must be set to 443.tls_port: 3478
Resources
The scalability of LiveKit is bound by CPU and bandwidth. We recommend running production setups on 10Gbps ethernet or faster.
When deploying to cloud providers, compute-optimized instance types are the most suitable for LiveKit.
If running in a Dockerized environment, host networking should be used for optimal performance.