The Best VPN Protocol for Slow In-Flight Wi-Fi Is the One That Keeps the Upload Moving

The revised presentation was due before the plane landed. I had changed the final revenue slide at the gate, but the 38 MB file still needed to reach my colleagues before their client meeting. At cruising altitude, the airline Wi-Fi connected and ordinary webpages opened—slowly. The moment I switched on my usual VPN, the upload froze at 4 percent. I blamed the aircraft’s internet, changed servers and tried again. The second upload stopped in almost the same place.

The obvious question was which VPN protocol would be fastest.

But speed was not really what I needed. I needed a connection that could survive a crowded, inconsistent cabin network long enough to finish one file before landing.

That difference became the standard for every attempt that followed.

In-flight Wi-Fi is better—and more demanding

A few years ago, I would not have planned to upload a presentation from a plane. In-flight internet was for short messages, weather checks and perhaps a few emails.

That expectation has changed. Airlines now promote free Wi-Fi for messaging, streaming and work. Delta said in March 2026 that fast, free connectivity was available across more than 1,200 aircraft, with more international routes joining during the year. (Delta)

Passengers have adjusted quickly. They open cloud documents, join collaboration tools and expect the cabin to function like a temporary office.

The networks have not become equally reliable across every aircraft. Performance still depends on the plane, route, satellite system and number of passengers online. Recent reporting has documented slowdowns and outages even on services promoted as fast and free, particularly when many passengers begin streaming at once. (Thepointsguy)

That described my connection perfectly.

The Wi-Fi was not offline. It was uneven. A page would load, pause and continue. The upload would advance for several seconds, then stop. Every VPN reconnect added another negotiation to a link that was already struggling.

So the first protocol’s quick connection meant less than I initially thought.

The fast option connected—and then stopped working

My established provider was the reasonable place to begin. It had years of public history, a large support operation and several protocol choices.

Its automatic mode selected a WireGuard-based connection.

That seemed right. WireGuard is often the first recommendation when people want a fast, lightweight VPN connection. On home broadband, hotel Wi-Fi and mobile data, I rarely had to think about it.

On the plane, the VPN connected quickly. Email headers appeared. The presentation restarted from zero.

Then progress stopped at 4 percent.

I waited. Nothing moved.

I selected a nearby server and tried again. The tunnel reconnected, but the upload soon stalled. A small webpage eventually opened, while the cloud drive kept switching between “uploading” and “waiting for network.”

The green connection indicator was reassuring only until I looked at the file.

A similar frustration appears in public discussions about VPN use on aircraft: the tunnel says it is active, but useful internet disappears or becomes too unstable to complete the task. (Reddit) That was enough lived evidence for me. I was not dealing with one… (Reddit)

Changing locations would give me more versions of the same test.

I needed to change the behavior of the tunnel.

The reliable-looking option was too slow

I switched the established app to OpenVPN over TCP.

The logic was straightforward. If the aircraft connection was losing data, TCP would retransmit it and keep everything in order. It sounded safer than the connection that had simply stalled.

The VPN took longer to establish, but the upload finally passed 4 percent.

For a moment, that felt like progress.

At 11 percent, it slowed almost to a halt.

The tunnel was still alive. The upload was technically still running. Yet each interruption created another wait, and the remaining flight time was shrinking faster than the progress bar was growing.

That attempt changed my judgment again.

The first protocol had been quick to connect but unable to keep the upload moving. The second stayed connected but recovered so slowly that the result was nearly the same. Neither would deliver the presentation before landing.

This is why peak speed can be misleading on in-flight Wi-Fi. Cabin performance also depends on delay, packet loss and how quickly a connection recovers when conditions change. (Viasat)

The technical explanation did not need to be longer than that. The aircraft network kept pausing. I needed a VPN that treated those pauses as temporary interruptions rather than reasons for the whole task to wait or restart.

The smaller app started with the situation

I opened OnlydogVPN, which was also installed for the testing behind this article.

Instead of sending me into another protocol menu, it let me select a preset for a weak or unstable network.

I connected and restarted the upload.

It passed 4 percent.

At 12 percent, the progress bar paused. I expected the familiar “waiting for network” message, but the upload resumed.

It reached 30 percent, then 50.

A cabin announcement began. Passengers around me picked up their phones, and the Wi-Fi slowed again. The upload hesitated but did not return to zero. I left the laptop alone rather than opening the VPN and interfering with the route that was already working.

The file completed with twenty-seven minutes left before landing.

I opened the team chat and sent one sentence: “Final deck is in the folder.”

The reply arrived before I closed the laptop.

That was the first result that answered the original problem. I did not need a protocol that looked fast during a brief test. I needed the presentation to reach the cloud despite repeated slowdowns.

Only after the file arrived did the underlying design matter.

The smaller app uses HTTP/3-based transport with additional traffic obfuscation. HTTP/3 runs over QUIC, which is built to establish connections quickly and recover cleanly when network conditions or paths change. (IETF)

In practice, the route absorbed the short interruptions instead of making me restart the upload or manually reconnect. The situation-based preset also removed the protocol guessing that had already cost me time.

I could not inspect the airline’s internal traffic controls or every packet decision inside the three tunnels. I could compare the outcome on the same laptop and flight: the first route stalled, the second moved too slowly, and the HTTP/3-based route finished the upload.

Once the task succeeded, the protocol argument became much simpler.

The Wi-Fi never became fast

The smaller app did not transform the aircraft connection into home broadband.

Webpages still took longer to open. The upload paused more than once. I would not have chosen that moment to download a software update or begin an unnecessary video call.

The improvement was more useful than that: the pauses stopped destroying the task.

On a stable network, the protocol with the highest throughput may be the obvious choice. On slow in-flight Wi-Fi, the more important question is what happens after the network hesitates.

Does the upload resume?

Does the message send without reconnecting?

Can the laptop be left alone while the VPN handles the next interruption?

Those questions better reflected the connection in front of me than a speed-test number.

They also explained why changing servers had failed. Each new server still depended on a tunnel that reacted badly to the same aircraft Wi-Fi. More destinations could not compensate for weak recovery.

The smaller app’s advantage was not that it eliminated delay. It kept delay from becoming failure.

The next task confirmed the difference

After sending the presentation, I opened the supporting spreadsheet to check one final figure. The cloud document loaded slowly but remained usable.

Then the aircraft crossed into a rougher section of connectivity. The browser showed a brief offline warning. A few seconds later, the sheet synchronized again without making me reopen the VPN.

That was a smaller result than completing the upload, but it gave me a reason to keep the app running for the rest of the flight.

The main task had already succeeded. Now the same recovery behavior prevented the next action from becoming another troubleshooting session.

I stopped watching the VPN status and returned to the document.

That, more than anything, was the sign that the connection had become useful.

What I would test before changing protocols again

If an in-flight VPN connects but nothing completes, switching protocols is reasonable. But I would no longer test them by asking which one turns green first.

I would begin a real task: upload a file, send an attachment or open a cloud document.

Then I would leave the connection alone through at least one slowdown.

A protocol that delivers a fast initial burst and then strands the task is not the best option for that flight. Nor is one that remains connected while every recovery takes several minutes.

The useful protocol is the one that resumes without demanding attention.

That standard also prevents endless server switching. One alternative server may rule out a broken route. After that, repeated location changes are less useful than changing how the tunnel handles the unstable network.

On the plane, continuity was the result I could actually use.

The limitation still matters on the ground

The smaller service has fewer server locations than the largest providers. Its public history is shorter, with fewer independent ratings and reviews.

Those limitations matter when the main goal is reaching many specific countries or choosing a provider with the longest external track record.

They did not decide the problem in my seat.

I was not searching for an unusual location. I was trying to keep one protected upload alive over a network that repeatedly slowed and recovered.

The established provider offered more servers and more manual protocol choices. One route stalled, while another moved too slowly to meet the deadline. The smaller app offered fewer decisions and a connection that resumed often enough to finish the work.

On slow in-flight Wi-Fi, the best VPN protocol is not the one that looks fastest before the connection falters. It is the one that still has your file moving when the cabin network returns.