Does enabling Mux always improve speed?

Not necessarily. Mux reduces handshake overhead by merging TCP connections, which offers noticeable improvement for applications that frequently establish many short-lived connections (like web browsing), reducing latency. However, for long-lived connection scenarios like large file downloads or video streaming, the improvement might be limited, and sometimes there might even be a slight overhead due to additional encapsulation headers. It's recommended to test and decide based on your actual usage scenario.

How do I choose between WebSocket and mKCP protocols?

The choice depends on primary needs and the network environment. If the network has interference or requires high obfuscation (e.g., to bypass firewalls), you should choose WebSocket + TLS. If the network is relatively open but extremely sensitive to latency and packet loss (e.g., online gaming, voice calls), and UDP is unimpeded, then mKCP is the better choice. For most daily browsing scenarios seeking a balance, WebSocket is a more universally stable solution.

How to verify the effect after performance tuning?

It is recommended to use a comparative testing method. First, record baseline performance (e.g., using `speedtest-cli` for speed tests, checking latency and routing via `ping` and `traceroute`, and noting the actual browsing experience). After modifying key parameters (like switching protocols, adjusting concurrency), perform the same tests under the same network environment and time period, and compare the data metrics and subjective experience. Using V2Ray's built-in stats function or logs to observe connection status and error messages is also an important method.

V2Ray Performance Tuning Guide: Best Practices for Configuration Parameters, Transport Protocols, and Network Environments

3/2/2026 · 3 min

Core Elements of V2Ray Performance Tuning

V2Ray, as a powerful network proxy tool, has its performance influenced by multiple factors including configuration parameters, transport protocols, and the network environment. Systematic tuning can significantly improve connection speed, stability, and resource utilization. Performance tuning is not about maximizing a single metric, but finding the optimal balance between speed, latency, security, and obfuscation.

Optimization of Key Configuration Parameters

Reasonable configuration is the foundation of performance optimization. The following parameters require special attention:

Concurrency: Adjust this in the streamSettings of outbounds. For high-bandwidth scenarios, appropriately increasing the concurrency (e.g., to 4-8) can utilize multiplexing to improve throughput. However, setting it too high may increase server load and connection establishment latency.
Buffer Size: Affects the processing capacity of TCP/UDP packets. In environments with significant network fluctuation or high latency (e.g., cross-border links), increasing the buffer size can reduce packet loss and retransmissions, improving smoothness. It's generally recommended to start with the default value and fine-tune based on actual conditions.
Mux Multiplexing: Enabling Mux (configured in the mux section of inbounds or outbounds) can merge multiple TCP connections, reducing handshake overhead. This is particularly suitable for scenarios like web browsing that require establishing many short-lived connections, effectively reducing latency.
Routing Strategy: Granular routing rules (e.g., direct connection for domestic traffic, proxy for specific traffic) can prevent unnecessary traffic from going through the proxy, reducing server load and improving speed when accessing domestic resources.

Selection of Transport Protocols and Flow Control Strategies

The choice of Transport protocol and underlying Stream Settings has a decisive impact on performance.

WebSocket (WS) / HTTP/2: Perform excellently in environments that need to bypass interference from middleboxes (e.g., certain corporate firewalls, ISP QoS). The WS protocol has great compatibility and is easily disguised as normal web traffic. HTTP/2 is more efficient but slightly more complex to configure. Both support TLS encryption for enhanced obfuscation.
mKCP: A UDP-based transport protocol that sacrifices some bandwidth for lower latency and stronger packet loss resistance. It is very suitable for latency-sensitive applications like gaming and real-time communication, but may be unavailable in networks that strictly restrict UDP.
TCP (Default): The most universal and stable transport method. In environments with good network quality and no special interference, plain TCP might offer the most reliable performance. It can be paired with different camouflage types (e.g., none, http, srtp) to counter Deep Packet Inspection (DPI).
Flow Control: V2Ray provides flow control methods like xtls-rprx-vision. The vision mode effectively resists active probing and, while ensuring security, improves connection establishment speed by reducing redundant data in the TLS handshake process.

Adapting to Different Network Environments

Tuning must consider the specific network conditions of both the client and server:

High Latency, Packet Loss-Prone Networks (e.g., long-distance cross-border links): Prioritize using the mKCP protocol and increase the buffer size. Enabling Mux also helps stabilize the connection.
High-Interference Networks (e.g., public Wi-Fi, campus networks): Use WebSocket + TLS or HTTP/2 + TLS for traffic obfuscation to avoid being identified and blocked by characteristic detection.
High-Speed, Low-Latency Networks (e.g., premium CN2 GIA lines): You can choose default TCP or HTTP/2, focusing on optimizing bufferSize and concurrency parameters to extract maximum bandwidth.
Mobile Networks: Frequent network switching suggests using protocols with fast reconnection capabilities (like WebSocket) and maintaining reasonable timeout settings.

Performance Monitoring and Iterative Tuning

Tuning is a dynamic process. It is recommended to use the v2ray stats function or third-party monitoring tools (like Prometheus + Grafana) to observe metrics such as traffic, connection count, and latency. By conducting A/B tests to compare the effects of different configuration combinations in real-world use and continuously iterating and optimizing, you can find the configuration scheme best suited to your current network conditions.