How to Choose a Shadowrocket Server: Latency Tests, Locations, and Protocols

Learn what Shadowrocket Connectivity Test results mean, how to compare servers by location and protocol, and why low latency doesn’t always mean faster page loads.

At a glance

This guide is for users who have imported a subscription into Shadowrocket and see multiple servers but aren’t sure which to choose. On the same network, use Connectivity Test to rule out options that repeatedly time out, then test your actual destination. Choose a consistently available server as your primary, and check a backup on another network you use regularly. Get subscriptions from your service provider.

Understanding Connectivity Test results

Run a test from Home → Connectivity Test on your iPhone or iPad. The number shown beside a server, in milliseconds, is how long that test took to complete. It is not a download speed or the time needed to load an entire webpage. Compare results only when the device, network, test method, and destination are the same. Don’t base your choice on an old reading after switching networks or before the test has finished.

1,000 ms
Equals 1 second; ms measures elapsed time
443/TCP
The default port for HTTPS over TCP, not the server’s port
UDP
Hysteria2 uses QUIC as its underlying transport

If the test only checks the TCP connection to the server, a successful result does not confirm proxy authentication or prove that the destination website is reachable. If the test sends a request to a specified URL through the server, the result also depends on that test destination and reflects only that request. A blank result does not mean 0 ms; a timeout means the test did not finish within the allotted time. One timeout alone is not enough to conclude that a server is offline.

Retest server candidates one by one from Home

  1. Keep the network consistent

    Choose the Wi-Fi or cellular network you use most, then wait for the network switch to complete before testing. Pause large downloads or file syncing during the test so temporary bandwidth use doesn’t look like server instability.

  2. Run the connectivity test

    Open Home → Connectivity Test and wait for the test to finish. Keep the method and destination unchanged, then run two more rounds. Note the time and number of timeouts for each round; don’t record only the lowest result.

  3. Keep two candidates

    Choose two servers that succeed consistently. If your use case requires a particular exit location, narrow the options using the location information from your service provider, then compare candidates in the same location. Don’t choose based on list ranking alone.

  4. Check real-world access

    On Home, select the first candidate and confirm the connection is enabled. Open the same webpage or perform the same action in the app. Confirm that the destination is routed through the proxy under your current configuration, then repeat with the second candidate. Don’t count an already-open tab or a page showing only cached content as a successful test.

  5. Record primary and backup

    Note the network, test time, and real-world results. If you usually connect in the evening, test again then. If you often leave Wi-Fi, check your backup separately on cellular.

For example, candidate A records 68 ms, 390 ms, and a timeout across three rounds; candidate B records 138 ms, 151 ms, and 145 ms. These figures illustrate how to compare results: A has a better best-case time, but B succeeds all three times with less variation. If both open the destination, check whether B can reliably complete tasks such as signing in, searching, or navigating pages. If B can’t open the destination, steady test results alone don’t make it a suitable primary.

Narrow your choices by location and protocol

The location in a server name is usually information provided by the service provider. It doesn’t necessarily reflect the physical distance from your phone to the server, nor does the name alone establish where a website sees the connection coming from. Peering, routing detours, and peak-time congestion can make a server in a nearby location slower. For location-dependent features, rely on your provider’s information and your actual results. If you don’t need a specific location, don’t discard a consistently reliable server just because its name suggests a farther one.

Location check

Location source
Provider label or description
Comparison conditions
Same device, network, and time of day
Verification
Can the destination website load and complete the required task?

Use location labels to filter; use real-world results to decide.

Protocol check

Type
Use the server type specified by your provider, such as VMess, VLESS, Trojan, or Hysteria2
Port
Use the value provided for that server; don’t infer it from the protocol name
Transport
Keep the parameters specified by your provider; Hysteria2 uses UDP-based QUIC

Compare the existing server entries; don’t change a server to a different protocol just to test it.

VMess and VLESS are different protocols. Whether TLS is enabled and which transport is used depend on the server configuration; you can’t infer them from the name. If several Hysteria2 servers repeatedly time out on the same network while an existing TCP-based server remains available, try the same Hysteria2 server on another network and consider a UDP path issue as one troubleshooting possibility. The two sets of entries may use different servers, so this alone does not prove that the network restricts UDP.

Choose a primary server for Wi-Fi and cellular separately

Track performance on both networks

Wi-Fi
  • Retest where you use the network most. If performance varies only when you’re far from the router, move closer and test the same server again.
  • If everything is slower at a particular time, wait and retest the original candidates. Don’t change both location and protocol at once.
Cellular
  • Make sure the device is no longer using Wi-Fi. Once the cellular connection is stable, compare the same set of candidates.
  • Run a quick check using the same webpage or app action and note which servers succeed consistently. When switching networks frequently, there’s no need to keep downloading large files.

If the same server is reliable on only one network, keep separate primary servers for Wi-Fi and cellular.

If the connection doesn’t stay active as expected after leaving Wi-Fi, first check the connection status on Home, then see whether Settings → On Demand has automatic connection conditions enabled for the current network. Confirm the connection status before blaming the server; otherwise, a drop during a network switch can be mistaken for a server timeout.

Low latency but slow pages: check routing, then identify the slow loading stage

The same website may take different routes depending on the mode selected in Home → Global Routing. In particular, under Config, switching servers doesn’t necessarily route the destination through the selected server. If the matching rule action is DIRECT, the website may connect directly before and after you switch servers, so its load time can’t be used to compare those servers.

Config

Recommended

Use the current configuration and its rules. Confirm that the policy matching the destination domain actually uses the server you’re testing. If it selects a different route, don’t include that run in the server comparison.

Best for: checking how a website or app performs under your usual rules

Direct

Establish a baseline using a destination that normally connects directly. If it’s slow over a direct connection too, check your local network or the website itself. Restore your original routing mode when you’re done.

Best for: determining whether slowness affects direct access or only the proxy route

Download Shadowrocket