VNPT Viettel FPT CMC Telecom Routing Comparison 2026

Somewhere off the coast of Vung Tau, a fishing boat drags its anchor across the sea floor and the AA-1 submarine cable goes dark. Within minutes, every ISP in Vietnam starts re-routing international traffic. The landing stations in Da Nang and Vung Tau keep working, but the path your packets take out of the country just changed. If you run a VPS for a business that touches Vietnam, this moment matters more than most people think. The carrier your VPS sits on decides whether your users see a 30 ms response or a 300 ms one.

The four big Vietnamese carriers, VNPT, Viettel, FPT Telecom and CMC Telecom, handle cable incidents very differently. Their international gateways, backup paths and domestic backbones are not equal. This post compares how they perform when a submarine cable fails, and what that means for a Vietnam VPS hosted on each network.

越南四大运营商在海底电缆中断时的回退路由差异很大，直接影响 VPS 的国际延迟。

The four major Vietnamese carriers differ significantly in failover routing during submarine cable incidents, which directly affects international latency for a VPS.

Why Vietnam ISPs rely so heavily on submarine cables

Vietnam's international connectivity is essentially a submarine cable story. The country has no land border with a major internet hub, it peers with the rest of the world through roughly a dozen undersea systems. The critical ones are the Asia-America Gateway (AAG), the Asia Pacific Gateway (APG), the Southeast Asia-Middle East-Western Europe 3 (SMW-3), the Intra-Asia (IA) cable, and the newer Vietnam-France-Hong Kong (VFH) line. Each carrier leases capacity across several of these, so when one fails they fall back to another. The problem is that the busiest routes, the ones carrying traffic to Hong Kong, Singapore and the US, all squeeze through the same narrow set of cables. When a single incident hits, say the AAG breaks near Vung Tau, every carrier feels it. But not equally. How much pain a customer feels depends on two things: how much spare capacity the carrier has on its other cables, and how smart its routing is at switching traffic without dropping sessions.

VNPT: the legacy incumbent with the largest cable portfolio

VNPT, the state-owned incumbent, has the most diverse cable portfolio of the four. It holds capacity on AAG, APG, SMW-3, IA and VFH. On paper that means the most backup paths. In practice, VNPT's international routing has historically been the slowest to converge after an incident. Partly that is scale, it carries more traffic than anyone else. Partly it is routing policy: VNPT tends to prefer AS-path length over latency, which can push traffic through longer but technically cheaper routes.

# Check which path your VPS takes to a Singapore endpoint
mtr -rwzc 50 -i 1 sg1.example.com

During the frequent AAG incidents of the 2010s, VNPT users reported the longest recovery times, often hours before international latency normalized. The carrier has improved since, but it still shows more latency variance during incidents than Viettel. For a VPS, VNPT is a reasonable default, but expect international traffic to degrade noticeably when a major cable breaks.

Viettel: the military-backed carrier with aggressive failover

Viettel is the most technically aggressive of the four. It built its own submarine cable landing station at Quy Nhon and holds capacity on APG, AAG and IA, plus terrestrial links into China and Laos. Viettel's routing team is known for aggressive BGP prefix tuning, and in recent incidents the carrier has consistently shown the fastest failover. Traffic shifts to alternate paths within minutes, sometimes seconds, of a cable fault.

# Compare latency to a US endpoint across carriers during an incident
ping -c 100 -i 0.2 8.8.8.8 | tail -n 3

The trade-off is that Viettel's international bandwidth per user is tighter. It has grown subscribers faster than its cable capacity in some years, so during peak hours international throughput can drop even without a cable fault. Do not read "fast failover" as "unlimited speed". For a VPS, Viettel's routing is the most stable during incidents, but you may see more throughput variance in the evening.

FPT Telecom: the commercial ISP with strong domestic peering

FPT Telecom focuses its investment on domestic fiber and datacenter connectivity. It operates a massive domestic backbone that links Hanoi, Da Nang and Ho Chi Minh City with multiple parallel paths. For international traffic, FPT leases capacity on AAG, APG and SMW-3, but it does not own landing stations. That means during incidents FPT depends on its upstream partners' failover, and it shows. In practice FPT's international latency during cable faults is mid-pack, better than VNPT, worse than Viettel. Where FPT excels is domestic performance. If your VPS serves users inside Vietnam, FPT's domestic routing is the most consistent of the four. The carrier also runs the country's biggest private datacenter ecosystem, so a VPS hosted at an FPT facility reaches other FPT customers over fast domestic links. For a VPS serving local Vietnamese traffic, FPT is a strong choice.

CMC Telecom: the enterprise-focused player with SmartCloud

CMC Telecom is the smallest of the four but targets the enterprise segment. It runs SmartCloud, one of Vietnam's few domestic public clouds, and holds capacity on APG and IA. During cable incidents, CMC's enterprise customers get priority treatment: the carrier's SLA structure means it reroutes business traffic first, often at the expense of consumer services. That is a meaningful advantage for a business VPS. The weakness is volume. CMC has less total international capacity, so a single cable fault consumes a larger share of its backup pool. Latency spikes may be sharper at the start of an incident, though recovery is usually quick because enterprise traffic is pushed to the front of the queue. For a VPS running business applications with users abroad, CMC's priority routing can be worth the trade-off.

Comparing failover behaviour in a table

How a cable incident actually affects your VPS traffic

A submarine cable break does not disconnect Vietnam from the internet. It changes the route. Traffic that went direct to Hong Kong on the AAG now goes through the APG, or through a transit provider in Singapore. Two things happen: latency rises because the path is longer, and packet loss appears because the backup links are congested with everyone else's traffic.

# Watch packet loss climb during an incident
watch -n 10 'ping -c 20 -i 0.5 1.1.1.1 | tail -n 2'

For a TCP-based service like a website or an API, this shows up as slow page loads and timeouts. For UDP-based services like video calls or VPNs, it shows up as jitter and drops. The practical consequence is that your VPS's carrier matters as much as its specs. A 2 GB RAM VPS on Viettel can feel faster during a cable crisis than a 4 GB one on a congested VNPT path. This is why the physical location of the VPS and the carrier's landing station matter. A VPS hosted in Ho Chi Minh City on Viettel's network reaches Viettel's Quy Nhon landing station over a domestic path that gets priority. The same VPS on a small reseller's network might depend on a single upstream that has one cable path and no backup.

What to check before renting a VPS in Vietnam

Before you rent a VPS, ask the provider three questions. First, which carrier is the upstream? Second, does the provider have links to more than one carrier for redundancy? Third, what was the latency and loss profile during the last cable incident? A provider that cannot answer these is not worth your money regardless of price.

# Test international path from a candidate VPS
curl -o /dev/null -s -w "connect: %{time_connect}s, total: %{time_total}s\n" https://www.google.com

A decent Vietnam VPS provider runs a BGP connection with multiple upstreams, often pairing Viettel and FPT, or Viettel and CMC. That way when one carrier's cable path degrades, the provider's router switches your prefix to the healthier one. This multi-carrier BGP setup is the single best insurance against cable incidents, but it costs the provider money, so only serious operations do it properly. At thueVPS, the Linux and Windows plans run on NVMe with a dedicated IPv4 in the Vietnam range, and the network is set up with redundancy across domestic carriers. That matters in the specific situation this post is about: a Linux VPS or a Windows VPS that keeps serving users even when a cable off the coast goes down.

How to test carrier performance yourself in 2026

Do not trust marketing pages. Run your own tests. Order the cheapest VPS from a provider, then measure international latency, domestic latency and throughput over a week. Track whether the numbers change during a cable incident. Vietnam sees several cable faults per year, so you will not wait long for real data.

# Install mtr on a Debian/Ubuntu VPS
sudo apt update && sudo apt install -y mtr-tiny iputils-ping

# Continuous trace to a Singapore and a US endpoint
mtr -rwzc 100 --curses sg1.example.com
mtr -rwzc 100 --curses us1.example.com

Look at the last hop in your carrier's network, that is the gateway to the international cable. Its latency to a Singapore peer should sit below 40 ms during normal operation. If it jumps to 80-120 ms when a cable is down, that is normal. If it jumps to 200 ms plus and stays there, the carrier is sending your traffic on a very long detour, and your VPS will feel slow for everyone outside Vietnam. Also test domestic latency between Hanoi, Da Nang and Ho Chi Minh City. A domestic RTT under 15 ms across the country is a sign of a strong backbone. Above 30 ms suggests the carrier routes domestic traffic through a single weak link, which does not bode well for an incident scenario.

What to do when a cable incident happens right now

When the news breaks that a cable is down, do not panic and do not switch providers immediately. First measure the current state of your VPS using the commands above. The failover may already have happened and your latency may be well within acceptable bounds. If it is not, check whether your provider's BGP has actually switched your prefix to another upstream.

# Check which upstream your VPS is reaching the internet through
ip route get 8.8.8.8

If the route looks wrong and your provider has alternative capacity, open a ticket asking them to shift your prefix. Many providers hold their failover policy at a certain threshold and will not trigger it for a single customer unless asked. Knowing this one trick often solves latency problems in minutes during an incident, instead of waiting hours for a global reroute.

FAQ

Related articles

• Vietnam domestic network performance VNPT Viettel FPT CMC
• Vietnam VPS vs Singapore VPS for serving Vietnamese users
• Why a Vietnam IPv4 matters when your users are in Vietnam
• Vietnam VPS vs offshore VPS for local traffic