For readers with a VLESS node who want to check their REALITY and Vision settings. REALITY handles the secure handshake between the client and proxy server; Vision handles part of the data forwarding after the handshake. Knowing which layer does what helps pinpoint whether mismatched fields like the address or public key are at fault, or whether the connection is established but limited by the network or application.
The REALITY Handshake: Node Address, SNI, and Target Site
A typical node configuration has four layers: VLESS is the proxy protocol, TCP (shown as RAW in some interfaces) handles transport, REALITY provides the security layer, and XTLS Vision is a VLESS flow-control mode. When a browser visits an HTTPS site, it also establishes a separate TLS session: the encrypted connection between the browser and the website. That session does not terminate at the proxy server or REALITY target site.
The client first connects to the address and port in the node configuration, then sends a TLS ClientHello containing the server name (SNI). The two sides use REALITY key material to validate the handshake, and the server checks handshake parameters such as the Short ID. Only after this layer succeeds does the server process the VLESS request and identify the account by user ID. In other words, an open port does not mean REALITY validation succeeded, and a successful REALITY handshake does not mean the VLESS user settings are correct.
- Node address: Determines where the network connection is made. SNI is a name field in the handshake; it does not redirect the TCP connection to another website.
- Target site: The TLS target configured on the server for REALITY. Handshakes that fail REALITY validation may be forwarded to this target, so seeing a response from the target site with a standard TLS tool does not prove the node can proxy traffic.
- Destination website: Determined by the application request. For example, when a browser visits a documentation site, the proxy forwards that request. The node's SNI does not redirect the web request to the site named in that field.
XTLS Vision: Which Encryption Work Does It Reduce?
Consider a browser downloading an HTTPS file. The browser has already wrapped the file transfer in TLS records. If the proxy connection wraps those records in another layer of TLS, the same payload incurs additional outer-layer encryption, decryption, and record processing. When Vision identifies inner TLS traffic that qualifies for passthrough, it can forward the relevant inner records directly over the proxy connection, reducing outer-layer processing for that traffic. TLS encryption between the browser and website remains in place, and the browser still validates the website's certificate.
- Initial phase: Proxy protocol processing takes place inside the outer secure connection while the inner handshake is inspected. Vision does not pass traffic through from the very first byte.
- When passthrough conditions are met: Subsequent data remains protected by inner TLS. The larger the file, the longer the sustained transfer—and the more opportunity there is to benefit from reduced outer-layer processing.
- When passthrough conditions are not met: Not all TLS traffic, or ordinary unencrypted application traffic, qualifies. Without a suitable inner TLS stream, do not expect the same processing path or performance gains.
So the “faster” in the title means Vision may reduce CPU overhead during sustained transfers; it does not guarantee faster page loads every time. For small pages, time to first byte often depends on DNS, TCP connection setup, the TLS handshake, and the website's response. Large-file throughput may instead be limited by the server's outbound bandwidth, network congestion, or your local connection speed.
v2rayN Setup Check: Flow, Public Key, and Short ID
After importing a subscription or share link, double-click the node in the v2rayN main window to open its settings and check each field. To add one manually, choose Add VLESS Server from the Server menu. Field groups and capitalization vary across versions; first confirm the type is VLESS, then check the connection fields and REALITY security fields separately.
VLESS Connection Settings
- Address
- Node hostname or IP address
- Port
- Port the server is listening on
- User ID
- VLESS user identifier assigned by the server
- Transport
- Select TCP or the corresponding RAW option as specified in the node configuration
- Flow
xtls-rprx-vision, if Vision is enabled for this user
The Flow setting must match the user's server-side configuration. Do not fill it in just because the security type is REALITY.
REALITY Security Settings
- Security type
reality- Server name / SNI
- Name allowed by the server; share links commonly use
sni - Public key
- Client-side value corresponding to the server's key pair; share links commonly use
pbk - Short ID
- Value allowed by the server; share links commonly use
sid - Fingerprint
- Choose the value specified in the node configuration; share links commonly use
fp
The public key is neither a website certificate nor the server's private key. The private key stays on the server.
The values below are for checking configuration fields, not benchmark results. Port 443 is just a common choice. A Short ID is written in hexadecimal; any non-empty value must contain an even number of characters, and leading zeros must be preserved. The 32-byte length refers to the raw X25519 key, not the character count of its encoded string in the interface.
Handshake Fails After Import: Check the Original Node Settings
Double-click the affected node and compare its port, SNI, public key, and Short ID. If several nodes in the same group stop working after a subscription update, first check whether the provider rotated the parameters. Do not substitute a website certificate for the public key, or change Flow to fix a failed security handshake.
How to Test: Separate Handshake Time, Throughput, and Resource Use
To compare Vision, configure two connections in advance on a server you are authorized to manage. Use the same core, transport, and security layer for both, enable Vision on one, and make sure each connection's Flow setting matches on the client and server. Clearing Flow on a client node that requires Vision only tests whether it can still connect; it cannot measure the performance difference.
- Keep the test path consistent: On the same device, use the same application to access the same HTTPS test resource. Keep the system proxy or TUN setup, routing rules, server, and test window consistent, and verify that requests actually use the node being tested.
- Measure the two phases separately: Use a new connection to measure time to first byte, then test throughput with the same file, large enough to reach sustained transfer. You can also record CPU usage for the Xray processes on both client and server during the transfer; do not use total system CPU usage as a substitute.
- Alternate and repeat: Run each test at least three times and compare the medians. If throughput is similar but Xray process usage drops, the benefit may mainly be lower resource overhead. If only one run shows a sudden throughput spike, retest before attributing it to Vision rather than network variation.
REALITY and Vision FAQ
Why is Flow blank when the subscription says REALITY?
REALITY and Vision are configured independently. First check whether the node provider enabled Vision for your user. If Flow is set in the original configuration but blank after import, check the protocol type in the settings window and the client version. If Flow is also blank in the original configuration, leave it as is.
Why does the latency test fail if the port is reachable?
A TCP port test only verifies that a connection can be opened. Retry the proxy request while checking the v2rayN core logs. If the connection times out, check the address, port, and network reachability. If the logs point to the REALITY handshake, verify the SNI, public key, and Short ID, and make sure the device clock is synchronized. If the connection is rejected after the handshake, check the VLESS user ID and Flow.
Will the same link work after switching to Android?
You can import it into v2rayNG, which uses the Xray core, then open the node settings to confirm that REALITY, Flow, the public key, and Short ID were retained. v2flyNG uses the V2Fly core; successful import alone does not mean it can run this Xray configuration.
Should I re-enter the public key if only one app won't open?
If the same node already proxies requests from other apps, leave the handshake settings as they are. Check whether the affected app respects the system proxy, or whether routing rules send it directly when TUN is enabled. Then use the logs to determine whether the issue is with DNS resolution, outbound selection, or the destination website's response.