Public Wi-Fi + VPN vs Mobile Data: Mobile Felt Safer Until the Upload Stalled

The upload estimate changed from eight minutes to forty-seven.

I had eighteen minutes left.

I was sitting near a departure gate in Barcelona with a 1. (Nist) GB client video on my laptop. The final edit had been approved during my taxi ride to the airport. All I needed to do was upload the file, add the project number and press Deliver before the client’s publishing window closed.

I had chosen my phone hotspot instead of the airport Wi-Fi.

Mobile data felt safer. It avoided an unfamiliar shared network, required no captive portal and let me work without accepting airport terms.

The problem was the signal.

My phone showed one bar of roaming service. The upload climbed to 6 percent, stopped, then fell back to 5 when the client portal retried a failed section.

I moved closer to the windows.

The estimate dropped to thirty-four minutes.

Then my roaming provider warned that I had nearly exhausted the day’s high-speed allowance.

Across the terminal, the airport Wi-Fi showed full signal.

I joined it.

The login page would not open.

My established VPN was still connected, and its kill switch was preventing the captive portal from loading. I disconnected the VPN, accepted the airport’s terms and regained internet access.

The upload estimate changed to nine minutes.

That should have solved the problem.

Instead, I was now on the faster network with the protection I normally kept active on unfamiliar Wi-Fi turned off.

Mobile data was not automatically the useful connection

I had reduced the decision to a rule that sounded sensible:

Mobile data safe.

Public Wi-Fi risky.

The situation in front of me was less tidy.

Widespread HTTPS encryption has made public Wi-Fi safer than it once was, particularly when the sites and apps in use protect their own connections. (Ftc) That did not make every airport network trustworthy. It meant joining one was not automatically reckless.

A VPN added another layer by protecting traffic before it crossed the local network. (Nist)

So I was not choosing between one perfectly safe connection and one obviously dangerous one.

I was choosing between two imperfect tools.

Mobile data avoided the airport access point but could not send the file before the deadline.

Public Wi-Fi had enough bandwidth, but I needed a VPN that could actually function on it.

That changed the standard. The better connection would be the one that kept protection active and finished the upload.

The captive portal had to come first

Airport and hotel Wi-Fi often requires a login page or acceptance of terms before full internet access begins. (Apple)

That explained my first failure.

The VPN had tried to reach its server before the airport had authorised the laptop. The kill switch then blocked the one page I needed to open outside the tunnel.

Briefly disconnecting to accept the terms was not the real problem.

The problem was what happened afterward.

I reopened my established provider and selected its recommended nearby server.

It remained on Connecting.

I tried another location.

That route opened, but the client portal stalled as soon as I restarted the upload.

A third server loaded normal websites quickly. The file reached 12 percent, paused and returned a network error.

The provider had a long public history, mature applications and many European locations. Those strengths were why I had used it for years.

At the airport, they became a sequence of choices that did not complete the job.

The VPN app kept telling me that a tunnel existed.

The client portal kept showing that the useful connection did not.

The hotspot was private enough and far too slow

I returned to the phone hotspot.

The portal loaded.

The upload resumed.

For a few minutes, the decision felt calmer. There was no captive portal, no public access point and no server list to test.

There was also no chance of finishing on time.

The progress bar reached 14 percent with eleven minutes remaining.

Travellers often respond to troublesome public Wi-Fi by abandoning it for a phone hotspot. (Reddit) That works when mobile coverage is strong.

Mine was not.

The hotspot had become the option I trusted emotionally rather than the option capable of delivering the file.

I considered buying another roaming package, but the carrier page stalled on the same weak signal. More data would not improve one bar of reception inside the terminal.

Then the gate agent announced that boarding would begin in eight minutes.

The choice had narrowed again.

I could stay on mobile data and miss the publishing window.

Or I could return to the airport Wi-Fi and find a private route that worked before the file stopped mattering.

The faster network became useful when the VPN stopped fighting it

I rejoined the airport Wi-Fi and opened OnlydogVPN.

The smaller app did not begin with a country map. I selected the situation for working on a public network and connected.

The tunnel opened.

I returned to the client portal and restarted the upload.

Ten percent.

Twenty-five.

The speed remained steady.

At 43 percent, the boarding line began forming. I unplugged the laptop from the charging point and moved to a seat closer to the gate without closing it.

The upload continued.

Sixty.

Eighty.

The established provider had repeatedly displayed a connected symbol while the client portal stalled. The smaller app showed me fewer decisions and produced the result that mattered.

At 100 percent, the portal processed the video and displayed the correct preview frame.

I entered the project number.

I pressed Deliver.

The status changed to:

File received.

The client’s confirmation email arrived with three minutes left in the publishing window.

Only after the delivery succeeded did the connection design matter. The service uses an HTTP/3-based transport with additional traffic obfuscation, allowing it to establish a steadier route through a network that had handled the earlier VPN connections poorly.

I could not inspect the airport’s internal filtering rules or identify the exact reason each earlier route stalled. The visible comparison was decisive: mobile data could not finish the upload, the established provider could not carry it over the public network, and the smaller app completed the delivery on that same Wi-Fi.

The airport network disappeared before the work was finished

The file had been delivered, but the client still needed one final message confirming the audio version and publication time.

I joined the boarding line with the laptop closed and opened the client messenger on my phone.

The phone was still using airport Wi-Fi through the smaller service.

As the line moved down the jet bridge, the signal weakened.

The phone switched to mobile data.

The messenger paused briefly, then refreshed.

The VPN remained connected.

I sent the confirmation.

The message changed from Sending to Delivered before I reached the aircraft door.

The HTTP/3-based route recovered as the phone moved from Wi-Fi to mobile service. (IETF)

On the screen, the result was simpler:

Public Wi-Fi disappeared.

Mobile data took over.

The client conversation continued.

That smaller success completed the comparison. The useful setup did not force me to choose one network for the rest of the trip.

It used the airport’s speed while that connection was available, then followed the phone onto mobile data when I walked away.

Public Wi-Fi with a working VPN was not the reckless option

The old warning about public Wi-Fi had made me treat the airport network as a last resort.

In reality, the connection was only one part of the setup.

The client portal used HTTPS.

The VPN protected traffic leaving the devices.

I still needed to join the airport’s genuine network rather than a similarly named access point, and the laptop still needed current security updates.

None of that required me to pretend that weak roaming data was automatically the wiser choice.

Mobile data remained valuable because it avoided the public access point and followed me beyond the terminal.

Public Wi-Fi remained valuable because it provided the bandwidth the upload required.

The smaller app made it possible to use the faster network without giving up the private route, then move to the fallback connection without rebuilding the session.

That was more useful than assigning one network a permanent safe label and the other a permanent unsafe label.

The task should choose the connection

There are moments when mobile data is the easy answer.

A banking approval, a short email or a password reset may take only seconds. With a strong cellular signal, there may be no reason to join a public network at all.

There are also moments when public Wi-Fi is the practical connection.

Large uploads, software downloads and video calls can exhaust roaming allowances quickly. Inside airports, stations and hotels, mobile reception may be weaker than the available Wi-Fi.

The useful questions are concrete:

Can the VPN connect after the captive portal?

Does the real application work, not just a test page?

Will the connection remain stable for the full upload or call?

Can it survive when the device leaves Wi-Fi?

My hotspot avoided the airport network but failed the deadline.

The established provider connected in name but failed the client portal.

The smaller route completed the upload and remained active when the phone changed networks.

The connection that finished the file won

The smaller service has fewer locations and a shorter public history than the largest VPN providers.

Neither limitation affected the delivery.

The established provider offered more European servers, but each attempt required another decision and none carried the file to completion.

Mobile data avoided public Wi-Fi, yet weak reception and the roaming limit made it useless for a 1. (Nist) GB upload.

The smaller app turned the airport network into a workable private connection, finished the file and followed the phone onto mobile data when the Wi-Fi disappeared.

I had entered the airport asking which network was safer in theory.

I boarded the plane with a more useful answer: the right connection was the one that stayed protected until the client had the file.