The same subscription can feel blazing fast for one person and painfully slow for another — the difference is usually not "having nodes" but knowing how to test and switch them. This guide makes it clear.
1. Latency is not speed
- Latency (Ping / ms): round-trip time. Low = responsive (web, gaming).
- Speed (bandwidth): how much data per second. High = smooth downloads/video.
A node can have low latency but choked bandwidth (peak hours) and still stall. Check both.
2. How to test latency
- Clash: in Proxies, tap the lightning icon (latency test) — each node shows ms, lower is better.
- Shadowrocket: on the home node list, tap latency test.
Latency tests hit a lightweight URL (e.g. gstatic.com/generate_204) — they reflect responsiveness, not real download speed.
3. How to test real speed (bandwidth)
After latency, connect to the node and verify with a real download: a speed-test site, a large file, or just your target use (play a video and see if it buffers). Rule of thumb: latency < 200ms and instant video = a good node; low latency but spinning video = bandwidth is choked, switch.
4. Let the client auto-select the fastest
- Clash: use a
url-testgroup to auto-pick the lowest-latency node (add afallbackgroup to switch on failure). - Shadowrocket: choose "Auto speed-test select".
proxy-groups:
- name: Auto
type: url-test
url: http://www.gstatic.com/generate_204
interval: 300
proxies: [A, B, C]
5. How to pick a node
- Closest wins: nearer = lower latency. For China content pick Hong Kong / Japan / Singapore; for overseas services pick the matching region.
- By purpose: streaming = unlock region; gaming = latency; browsing = any low-latency node.
- Keep spares for peak hours: 20:00–23:00 is most congested — keep 3–5 nodes to switch instantly.
6. How to switch nodes
- Clash: tap the node in Proxies / a policy group; url-test switches automatically.
- Shadowrocket: tap a node on the home screen; reconnect once so the new node takes effect.
7. Still slow after switching? It may be your network, not the node
If all your nodes slow down or drop at once no matter how you switch, the problem is usually your local network, not the subscription:
- ISP QoS / line throttling: carriers (China Mobile / Unicom / Telecom) sometimes throttle or physically cut international bandwidth, especially at peak hours or sensitive periods — every line stalls together and switching nodes doesn't help.
- How to tell: one node slow → switch nodes; all nodes slow/down while domestic sites are fine → the international gateway is being throttled, not the node.
- What to do: change protocol / port (some are harder to fingerprint), change network (ISP, Wi-Fi vs cellular), avoid peak hours, keep several lines with different exit locations; throttling often eases after a while.
Key takeaway: no subscription can fully beat physical-layer throttling — but multiple nodes, multiple protocols and easy switching minimize the impact.
8. The prerequisite: a multi-node subscription
Testing, auto-select and switching all require a subscription with multiple nodes. SSRocket provides a multi-node subscription + one-tap import: after payment your config is emailed automatically; import into Clash / Shadowrocket and you get several nodes to test and switch freely, with backups for peak hours.
FAQ
- Low latency but still slow? Latency ≠ bandwidth — the node is congested at peak, switch or use auto-select.
- All nodes suddenly slow/down? Usually ISP throttling of the international line — change protocol/port, change network, avoid peak, keep spare lines.
- Which node should I pick? Closest + by purpose; when unsure, enable auto speed-test select / url-test.