Measuring real latency from Ho Chi Minh City and Hanoi to your server

Latency to or from Vietnam depends on which Vietnamese ISP the traffic passes through, the route your provider takes to reach Vietnam, and your own domestic network. You need a reproducible method to measure, not gut feel. This post gives you the tools, mtr, ping, traceroute, public looking glasses, third-party probes, to run your own latency vietnam server test from Ho Chi Minh City, Hanoi, or any Vietnam point of presence. No guesswork, just commands and real output interpretation.

• Key takeaways
• Use mtr --report for a combined live traceroute and packet-loss view per hop.
• Vietnam has four major ISPs (Viettel, VNPT, FPT, CMC Telecom); each can produce different latency to the same target.
• Public looking glasses let you initiate a ping or traceroute from inside a Vietnam ISP without owning a server there.
• A Vietnam VPS at a provider like thueVPS gives you a test endpoint within the Vietnam IP range, useful for running your own probes.
• Do not rely on a single measurement; test across different hours and different ISPs to build a representative picture.

用mtr和traceroute自己测河内与胡志明市的延迟，不要只看宣传数字。

Measure latency from Hanoi and Ho Chi Minh City yourself with mtr and traceroute instead of trusting a marketing number.

Why latency from Vietnam matters

Your server may run a web application, an API, a game server, or a real-time service that Vietnamese users need to reach. If the round-trip time (RTT) from a Vietnamese user to your server exceeds ~200-250 ms, the experience degrades, page loads stall, interactive apps feel sluggish. Every 100 ms of additional latency costs you conversions, engagement, or user satisfaction.

The critical insight: latency is not a single number. It varies by which Vietnamese ISP the traffic traverses (Viettel, VNPT, FPT broadband, CMC Telecom), by time of day (peak evening hours vs. late night), and by the path your traffic takes to leave the country. A measurement from one VPS inside Vietnam that uses only Viettel, for example, does not tell you the latency a user on FPT broadband will see.

You also need to understand that any hop leaving Vietnam crosses international transit, the result depends on the other side's domestic network too. Do not expect a magic pipe. If your server sits in the United States, for instance, the geographic distance (roughly 13,000 km) imposes a physics floor of about 130 ms RTT, even with optimal routing. The real number will be higher.

What you need before you start

• A target server or IP address you want to test against.
• A machine on which you can run ping, traceroute, and mtr. Any Linux or macOS system works. Windows users need WinMTR or pathping.
• (Optional) A VPS inside Vietnam that you can SSH into and run commands from. For Vietnam location tests, a VPS with a Vietnam IP range from providers like thueVPS gives you a local endpoint.
• The URLs of public looking glasses operated by Vietnamese ISPs (listed below).

Tool 1: ping, the baseline

ping sends ICMP echo requests and measures RTT. It is the simplest latency test but tells you only the end-to-end value, not where any loss or delay occurs.

ping -c 20 your-server-ip

The -c 20 flag sends 20 packets. Output looks like:

--- your-server-ip ping statistics ---
20 packets transmitted, 20 received, 0% packet loss
rtt min/avg/max/mdev = 245.123/260.789/301.456/15.234 ms

What to look for: average RTT and packet loss percentage. If loss exceeds 1% on a 20-packet run over a reasonable connection, the path likely has a congestion or routing problem.

Ping from inside Vietnam: SSH into your Vietnam VPS and run the same command toward your server. That gives you the round trip from inside Vietnam to your target. Reverse the direction, ping the Vietnam VPS from your server, to see the opposite path.

Tool 2: traceroute, the path

traceroute maps each intermediate hop between you and the target. It reveals where traffic enters and leaves Vietnam, and which ISP segments are the bottleneck.

traceroute -n your-server-ip

The -n flag skips DNS lookup (faster, no hostname noise). Example output:

 1  10.0.0.1    0.5 ms
 2  172.16.0.1  1.2 ms
 3  203.113.0.x 5.3 ms   [Viettel gateway]
 4  203.113.1.x 8.7 ms
 5  59.153.0.y  12.1 ms  [international transit]
 6  72.14.238.a 60.2 ms  [Google transit]
 7  216.58.200.b 120 ms  [target]

Key observations:

• Hops 3 through 5 show the Vietnam domestic segment. A large jump in RTT between hop 2 and hop 5 (e.g., from 5 ms to 60 ms) is normal, it marks the exit from Vietnam.
• Any hop showing * * * for more than one probe means that router does not respond to ICMP, which is common. Do not interpret it as a failure.
• A hop with RTT significantly higher than the next hop (e.g., 300 ms then 220 ms) indicates pacing or congestion on that router.

Run traceroute from inside Vietnam and from your own location. The differences tell you where the asymmetry is.

Tool 3: mtr, the gold standard

mtr combines ping and traceroute into a single continuous measurement. It sends probes to each hop iteratively and updates the statistics in real time. For a definitive latency vietnam server test, this is your primary tool.

mtr --report --report-cycles 10 your-server-ip

Flags explained: --report runs a fixed number of cycles and prints a table at the end. --report-cycles 10 sends 10 packets to each hop.

Example output snippet:

                               Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 10.0.0.1                   0.0%    10    0.3   0.4   0.2   0.8   0.1
 2. 172.16.0.1                 0.0%    10    1.1   1.3   0.9   2.1   0.3
 3. 203.113.0.x                0.0%    10    5.2   5.8   4.9   8.3   0.6
 4. 203.113.1.x                0.0%    10    8.1   8.9   7.8   12.0  1.1
 5. 59.153.0.y                 0.0%    10    58.2  62.4  55.1  71.3  4.5
 6. 72.14.238.a                0.0%    10    119.3 122.1 117.8 131.0 4.2
 7. 216.58.200.b               0.0%    10    199.5 201.2 198.3 205.4 2.1

Read the Loss% column per hop. If you see loss on an intermediate hop but no loss at the final hop, the router is rate-limiting ICMP, ignore it. If you see loss on the final hop while earlier hops show zero loss, the target is rate-limiting. If loss accumulates along the path (e.g., 5% on hop 3, 10% on hop 4, 15% on hop 5), you have a real congestion issue on that segment.

Run mtr from multiple sources: your own connection, a Vietnam VPS using Viettel, one using FPT, and a looking glass inside Vietnam. Compare the average column at the final hop.

Public looking glasses for Vietnam ISPs

A looking glass is a web interface provided by a network operator that lets you run ping, traceroute, and sometimes BGP queries from inside their network. For Vietnam, the main ISPs operate public looking glasses:

• Viettel: lg.viettel.vn (or check their current looking glass page, URLs change). Viettel is the largest ISP in Vietnam, covering both mobile and fixed broadband.
• VNPT: VNPT (Vietnam Posts and Telecommunications Group) operates a looking glass at lg.vnpt.vn. VNPT owns the legacy copper and fiber infrastructure; their routing often differs from Viettel's.
• FPT Telecom: lg.fpt.vn. FPT runs its own international gateways and peers at major IXPs. FPT is the second-largest fixed broadband ISP.
• CMC Telecom: lg.cmctelecom.vn. CMC focuses on enterprise and datacenter connectivity; their looking glass is less widely known but useful for corporate-NIC testing.

How to use them:

1. Open the looking glass URL in a browser.
2. Enter your server's IP address as the target.
3. Select ping or traceroute as the test.
4. Run it. The results show you the latency from that ISP's infrastructure to your server.

Because each ISP may route differently, run the same test across all four. A difference of 30 to 80 ms between them is common, one ISP may peer directly with your provider, while another sends traffic through a congested exchange.

Third-party probe networks

If you do not have a Vietnam VPS and want measurements from multiple Vietnam locations without maintaining SSH access, use a third-party probe network. Two options:

• Cloudflare Observatory: Cloudflare's speed test runs from its edge nodes. If you have a site behind Cloudflare, the Observatory report shows request latency per region, including Vietnam.
• Catchpoint or ThousandEyes: These commercial platforms have probes inside Vietnam (Viettel, VNPT, FPT). They are not free, but you can request a trial to run a one-off latency vietnam server test from Vietnam PoPs. The output includes per-hop loss and jitter.
• Speedtest by Ookla: Not directly latency, but you can run a speedtest from a Vietnam server to your target if you set up an Ookla speedtest server on your machine. The Ookla client shows ping times from each test. This is a rough indicator, not a replacement for mtr.

Interpreting results: what is good latency from Vietnam?

Latency from Vietnam to common international destinations (measured from a typical Viettel residential mobile connection to cloud providers):

If your measurement falls well outside these ranges (e.g., 400 ms to Singapore), something is wrong, either a routing detour (traffic going through the US or Europe first) or congestion on an intermediate link. Investigate the path using mtr hop by hop.

For serving Vietnamese users, latency below 50 ms is typical for a server inside Vietnam. That is the primary reason to host your application locally, the domestic round-trip between the user and a Vietnam datacenter stays under 10-20 ms for most ISPs. For applications with Vietnamese users, the fastest path is a server that terminates inside the country.

Practical workflow for a thorough latency test

Do not stop at one ping. Follow this sequence:

1. Ping from your local machine to your server. Record the average RTT.
2. Ping from a Vietnam VPS (if you have access) to your server. Record the average RTT.
3. Run mtr --report from the Vietnam VPS to your server. Note which hop crosses the international gateway and the RTT jump at that hop.
4. Run traceroute from each Vietnam ISPs' looking glass to your server. Compare the domestic RTT per hop.
5. Run the same tests during Vietnam peak evening (20:00-22:00 ICT) and during low-use hours (03:00-05:00 ICT). Latency can vary by 20-50 ms. If the peak-time path shows 10%+ loss at any hop, that ISP segment is oversubscribed.
6. Document everything: keep the mtr reports, note the ISP and approximate location (if you know it). When troubleshooting with your provider, having an mtr report from inside Vietnam is the most actionable data you can give them.

What about jitter?

Latency variability (jitter) matters for real-time applications like VoIP, video calls, and online gaming. The StDev column in mtr's report output shows the standard deviation of RTT per hop. A StDev above 10 ms on the final hop indicates significant jitter, the path is either congested (bufferbloat) or using wireless links with variable delay.

Run mtr in live mode (mtr your-server-ip, no --report) and watch the Last column bounce. If it swings 30+ ms per packet, you have a jitter issue. A jitter buffer can mask small variations, but large swings degrade interactive applications and cause retransmissions, effectively lowering throughput.

FAQ

Related articles

• Vietnam VPS vs Singapore VPS for serving Vietnamese users
• Why a Vietnam IPv4 matters when your users are in Vietnam
• Should your CDN origin sit inside Vietnam
• Do you need a Vietnamese entity to buy hosting in Vietnam