Moscow to China Route Test: Identify CN2, Measure Latency and Packet Loss
Routes from Moscow servers to mainland China vary widely: some detour through European nodes before heading back to China, while others connect directly via networks such as CN2. Different routes mean clearly different latency, evening-peak packet loss and download speeds for users in China.
This guide shows you how to use common command-line tools to test a Moscow server across five dimensions, outbound route, return route, latency, packet loss and bandwidth, and determine whether the route is CN2 and whether it takes a detour.
Key Takeaways
- Routes from Moscow servers to mainland China vary widely: some detour through European nodes while others connect directly via networks such as CN2, which leads to clearly different latency, evening-peak packet loss and download speeds.
- Hops in AS4809 or the 59.43.x.x range show that a route passes through China Telecom's CN2 network, and a route with 59.43.x.x throughout and no 202.97.x.x is usually CN2 GIA.
- Latency from mainland China to Moscow is typically around 100–160 ms over a direct path and typically above 200 ms when the route detours through Europe, but your own test results are what count.
- To test a Moscow server's route, use ping for latency and packet loss, mtr and NextTrace for the return route, and iperf3 for real bandwidth, both during the day and at evening peak.
- IMIDC's Moscow data center offers CN2 routes direct to mainland China, and you can request a test IP to run these tests yourself before deciding.
Requirements / Preparation
- A Moscow server (or a test IP from your provider) running Debian, Ubuntu, Rocky Linux, AlmaLinux or similar, with root access. For how to log in, see Connecting to a Linux Server via SSH.
- A computer in mainland China, ideally able to test from China Telecom, China Unicom and China Mobile networks separately.
- Two concepts to understand: the outbound route goes from your local machine to the server, and the return route goes from the server back to you. Page loads and downloads are mainly affected by the return route.
1. What to Look For: Route Signatures of CN2 and Detours
| Route signature | Meaning |
|---|---|
| Hops in AS4809 or the 59.43.x.x range | Passes through China Telecom's CN2 network |
| 59.43.x.x throughout, no 202.97.x.x | Usually CN2 GIA |
| 202.97.x.x (AS4134) on the domestic segment, 59.43.x.x after leaving China | Usually CN2 GT |
| Passes through European city nodes such as Frankfurt, Amsterdam or Stockholm | Detours through Europe back to China |
| Hops go straight from Moscow into China's domestic backbone | Direct path |
| AS58453 / AS58807 | China Mobile CMI / CMIN2 |
| AS4837 / AS10099 / AS9929 | China Unicom networks |
For more ways to identify routes, see: Server Route Testing Tutorial: Identifying CN2 GIA and Return Routes.
2. Test Latency and Packet Loss With ping
On a computer in China, run:
# macOS / Linux
ping -c 100 SERVER_IP
# Windows
ping -n 100 SERVER_IP
Focus on three values:
- packet loss: the packet loss rate; lower is better;
- avg: average latency;
- mdev / difference between max and min: jitter; smaller means more stable.
From mainland China to Moscow, latency over a direct path is typically around 100–160 ms, while routes that detour through Europe are typically above 200 ms. These are just common industry ranges; actual results depend on your province, carrier and time of day, so go by your own test results. We recommend testing once during the day and once during evening peak hours (20:00–23:00 China time).
3. Check Per-Hop Packet Loss With mtr
Install mtr on the server and test the return route to an IP in China:
# Debian / Ubuntu
apt update && apt install -y mtr-tiny
# Rocky / AlmaLinux
dnf install -y mtr
# Send 100 packets to the target IP in China and show AS numbers
mtr -rwzc 100 CHINA_TARGET_IP
Options: -r report mode, -w show full hostnames, -z show AS numbers, -c 100 send 100 packets.
How to read the results:
- Only pay attention to packet loss that starts at a hop and continues all the way to the destination. Loss at a single intermediate hop with normal hops after it is usually just a router rate-limiting ICMP;
- Check the AS column for AS4809 (CN2) or AS4134 (163);
- Check city identifiers in hostnames to see whether the route passes through European nodes.
You can also use mtr on your computer in China to test the outbound route (install it via Homebrew on macOS, or use WinMTR on Windows).
4. Visualize the Route With NextTrace
NextTrace automatically labels each hop with its ASN, carrier and geographic location, making it very easy to see whether traffic detours through Europe. Before installing, check the official repository (github.com/nxtrace) for the latest installation method. The common one-line install is:
curl nxtrace.org/nt | bash
Test the return route:
nexttrace CHINA_TARGET_IP
If European cities such as Frankfurt, Amsterdam or Stockholm appear in the route, traffic is relayed through Europe. If China nodes and AS4809 hops appear right after Moscow, the route is a direct CN2 path.
Tip: when testing the return route, pick test IPs in China on China Telecom, China Unicom and China Mobile, and run the test once for each carrier.
5. Measure Real Bandwidth With iperf3
Beyond latency and packet loss, real throughput matters just as much. Start the server side:
apt install -y iperf3 # or dnf install -y iperf3
iperf3 -s
On your computer in China, run:
# Test the outbound route (upload from local to server)
iperf3 -c SERVER_IP -t 30 -P 4
# Test the return route (download from server to local); add -R to reverse
iperf3 -c SERVER_IP -t 30 -P 4 -R
-t 30 runs for 30 seconds and -P 4 uses 4 parallel streams. iperf3 uses port 5201 by default, so make sure your firewall allows it before testing. When you're done, press Ctrl+C to stop the server so the port isn't left exposed.
6. How to Judge the Results
| Test | Ideal result | Warning sign |
|---|---|---|
| Return route | AS4809 / 59.43 hops, no detour through Europe | Relayed through multiple European nodes |
| Evening peak packet loss | Close to 0% | Sustained loss above 1–2% |
| Latency jitter | Small gap between average and maximum | Large latency swings at evening peak |
| iperf3 return route | Close to the bandwidth you purchased | Far below purchased bandwidth with large fluctuations |
If performance drops noticeably at evening peak, take screenshots of your mtr and NextTrace results and open a ticket with your provider for investigation.
FAQ
If I see 59.43 hops, is it definitely CN2 GIA?
Not necessarily. 59.43.x.x means the route passes through the CN2 network, but if 202.97.x.x also appears on the domestic segment, it's usually CN2 GT. You need to look at the full route to judge.
Why are the outbound and return routes different?
Internet routing is asymmetric. Outbound and return traffic may pass through different carriers and nodes, so you need to test both directions.
Is it normal for China Unicom and China Mobile users not to see CN2?
Yes. CN2 is a China Telecom network, so routes for China Unicom and China Mobile users depend on how the carriers interconnect. Judge by actual latency and packet loss.
If ping fails, is the server down?
Not necessarily. Some servers or firewalls block ping. You can use tcping to test a TCP port instead; see the Network Troubleshooting Guide.
Conclusion: Let the Data Decide
The only way to know if a route is good is to test it. Use ping for latency and packet loss, mtr and NextTrace to see whether the route uses CN2 or detours through Europe, and iperf3 for real bandwidth. This method works for any overseas server.
IMIDC's Moscow data center offers CN2 routes direct to mainland China. Visit the IMIDC website to request a test IP, run the tests in this guide yourself, and then decide. Our Chinese- and Russian-speaking technical team is online 24/7 and can help you analyze your results.