The Internet didn’t break; we just forgot what it was leaning on

In the days after the Bangladesh Internet shutdown began on 18 July 2024, I was at home, and nothing worked. Messages wouldn't send. Websites wouldn't open. Every app I normally relied on just sat there doing nothing. The only thing that still worked was an ordinary phone call, so that's what I used. But my business had stopped. I couldn't reach my clients, and I had no way to tell them why.

I thought about those days again when I read the INNOG 9 report from APNIC 62. One of the talks, by Sumon Ahmed Sabir of Fibre@Home, was about exactly this: What happens when local networks keep running but the links to the outside world disappear?

You can read Sumon's slides yourself. Here is what stuck with me, and why I think it matters for Bangladesh and the rest of South Asia.

What happened to connectivity in July 2024

It's worth being precise here because people mix this up. The July 2024 blackout was not caused by a cable failure. Internet Society Pulse and Open Observatory of Network Interference (OONI) reported that the disruption began on 18 July amid restrictions on Internet access.

According to Internet Society Pulse, connectivity across the economy was almost completely gone by that afternoon and began returning slowly after about five days. OONI reported that the full blackout lasted from 18 to 23 July. Social media and messaging apps were then blocked from 24 to 31 July.

The damage was significant, although estimates vary. The e-Commerce Association of Bangladesh (e-CAB) put losses to the sector at around BDT 20 billion (Tk 2,000 crore). Internet Society's NetLoss calculator estimated losses of more than USD 21 million in the first five days. These are estimates, not audited figures, but they point in the same direction - losing connectivity has substantial economic consequences.

The part nobody thinks about

For network operators, the reason international connectivity disappears may vary. The operational question remains the same: What continues to work locally when the outside world can't be reached?

Sumon highlighted how services that could otherwise continue operating locally may fail because they depend on infrastructure outside the economy.

His examples included Domain Name System (DNS) resolvers hosted abroad, cloud-based application components, remote Application Programming Interfaces (APIs), time synchronization, and mobile payment platforms. On a normal day, you never notice these dependencies. During an outage, any one of them can stop a service that could otherwise have continued operating locally.

For me, it felt like everything went down at once. I couldn't tell which of my tools had failed because the network was unavailable and which had failed because they depended on something outside the economy.

That is the question Sumon wants operators to answer before the next outage, not during it. His recommendations focused on reducing these external dependencies, including strengthening local DNS infrastructure and planning for extended outages.

The broader lesson for operators is simple: Identify which locally delivered services still rely on infrastructure outside the economy, then consider how those dependencies can be reduced.

Bangladesh's international connectivity

Bangladesh's international connectivity is concentrated across a relatively small number of submarine and terrestrial routes. SEA-ME-WE 4 and SEA-ME-WE 5 provide submarine connectivity, alongside terrestrial connectivity through India.

Routine maintenance also shows why route diversity matters. During repair work on SEA-ME-WE 5 in April 2026, users experienced slower connectivity and some interruptions, while SEA-ME-WE 4 remained in service.

Demand for bandwidth continues to grow, and additional international connectivity is planned, including SEA-ME-WE 6. More capacity and additional routes can improve resilience. But they do not remove the need to understand the external dependencies of services intended to operate locally.

It's not only our story

India provided another example during the same session. Ashutosh Zutshi of NEC India highlighted the concentration of international connectivity at a relatively small number of landing locations, particularly Mumbai and Chennai.

India and Bangladesh differ considerably in scale. But both examples show why geographic and network diversity in international connectivity matter alongside overall capacity.

What operators can do now?

The discussion suggests several practical steps operators can take without waiting for additional international connectivity:

• Identify services that depend on infrastructure outside the economy.
• Deploy appropriate local DNS infrastructure and test how services behave when international connectivity is unavailable.
• Identify critical content and services that could be made available locally.
• Test continuity arrangements against extended disruptions, not just brief ones.

One last thought

We often think about resilience in terms of available capacity. Sumon's talk suggested another useful measure: How much of a network continues to work when international connectivity is unavailable?

Bangladesh's experience shows why that question matters. The lessons are relevant well beyond its borders.

Imteaz Chowdhury is a writer based in Dhaka, Bangladesh. He writes about domain names, Internet policy, and digital infrastructure on CircleID. He worked through the July 2024 Internet shutdown, which is where this article began.

The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.