Add a UDP listener to an Edge Load Balancer (ELB) instance to forward UDP traffic to backend servers. UDP listeners suit latency-sensitive workloads—such as video chat and real-time financial data feeds—where speed matters more than guaranteed delivery.
Prerequisites
Before you begin, ensure that you have:
An ELB instance. See Create an ELB instance.
UDP listeners do not support ports 250, 4789, or 4790. These ports are reserved by the system.
Add a UDP listener
Log on to the ENS console.
On the ELB Instance page, find the ELB instance you want to configure, and click Listener Configuration in the Actions column.
On the Listener tab, click Add Listener.
Configure the listener parameters.
Parameter Description Example Listener name A display name for the listener. udp-listener-prodELB protocol The protocol for this listener. Select UDP. UDP Listener port The port on which ELB receives incoming requests and forwards them to backend servers. Valid values: 1–65535. Ports 250, 4789, and 4790 are not available. 53(Optional) To configure advanced settings, click Modify under Advanced Settings.
Parameter Description Scheduling algorithm The algorithm used to distribute traffic across backend servers. See Choose a scheduling algorithm. Obtain client IP address Whether ELB passes the original client IP address to backend servers. When enabled, backend servers receive the real client IP. Supported on Layer 4 listeners only. Enabled by default. Click Next and configure health checks. See Configure health checks.
Click Next, review the settings, and click Create.
On the Listener tab, find the new listener and click Enable in the Actions column.
Choose a scheduling algorithm
Select the algorithm that best matches your traffic pattern and backend setup.
| Algorithm | How it works | Best for |
|---|---|---|
| Weighted Round-Robin (WRR) | Distributes requests in rotation, weighted by server capacity. Higher-weight servers receive proportionally more traffic. | General-purpose workloads with servers of different capacities. |
| Weighted Least Connections (WLC) | Routes each new request to the server with the fewest active connections, adjusted for server weight. If multiple backend servers have the same weight, requests are forwarded to the backend server that has the least number of connections. | Long-lived connections or workloads with variable request duration. |
| Round-Robin (RR) | Distributes requests evenly to all backend servers in sequence, regardless of weight. | Uniform workloads where all backend servers have equal capacity. |
| Consistent Hashing (CH) | Routes requests from the same source to the same backend server, using either source IP or Quick UDP Internet Connections (QUIC) ID as the hash key. | Stateful workloads that require session affinity. |
| IQCH | Routes requests to the same backend server using consistent hashing on specific three bytes of the iQUIC connection ID from the second byte. | Workloads using the iQUIC protocol that require connection-level affinity. |