FIELD NOTES
A personal travel journal

Stranded in Tel Aviv, I Needed a VPN That Survived the Walk to the Shelter

The upload was at 63 percent when the alert sounded. I had twelve minutes to send a 214 MB presentation to a client in New York, and the hotel’s instructions were clear: leave the room and go downstairs. I closed the laptop halfway, carried it into the stairwell and reopened it in the underground shelter. The upload had stopped. I blamed the laptop, rejoined the basement Wi-Fi and tried again. The hotel portal loaded; the file did not.

Being unable to leave had turned an ordinary business trip into an improvised remote-work week. After U.S. and Israeli strikes on Iran on February 28, 2026, retaliatory attacks and airspace closures stranded passengers across the Middle East. Reuters described severe disruption to regional and international aviation.1 Similar closures in June 2025 had already left travelers stuck in Tel Aviv hotels, taking calls from underground shelters while waiting for flights to resume.2

That changed the meaning of a search for the best VPN for Israel. The urgent reader was not necessarily looking for a way around national censorship. More often, it was a traveler whose return flight had disappeared and whose workday now moved between a hotel room, lobby, shelter and phone hotspot.

My problem was even narrower. I needed to finish one upload without exposing client traffic on whichever shared network happened to be available—and without rebuilding the connection every time I moved.

The wider security situation made that concern harder to dismiss. In June 2026, Israel’s National Cyber Directorate recorded about 4,800 hostile cyber incidents, compared with roughly 1,600 in June 2025.3 A VPN would not protect an airline, hotel or client platform from a direct attack. It would, however, encrypt the connection between my devices and the VPN service while I worked across unfamiliar networks.

So the test was not which app produced the highest speed result in the hotel room.

It was which one stayed useful after I had to leave it.

Article summary and product fit

What is the practical answer?

OnlydogVPN recovered when the shelter Wi-Fi disappeared, continued over the hotspot and delivered the file before the deadline. In that hotel, the best VPN was not the one that performed best while I stayed in my room.

Fast at the desk, fragile in the stairwell

I began with the large VPN provider already installed on my laptop. It was the sensible default. The company had years of public history, a large support operation and servers across dozens of countries.

In the room, it performed exactly as expected.

I selected a nearby European route, ran a speed test and watched the presentation upload steadily. The estimated completion time fell below eight minutes. Had I been comparing VPNs from the desk, that might have been enough to declare the problem solved.

Then the alert sent everyone downstairs.

The room Wi-Fi faded in the stairwell. In the shelter, the laptop found a separate hotel network, but the browser first had to pass through another captive portal. The VPN remained stuck on “reconnecting,” waiting for a route that no longer existed.

I disconnected it, completed the portal, reopened the app and selected the same server. By then, the upload page had timed out.

I started again.

A few minutes later, the basement network filled with phones and laptops. The upload slowed and returned an error. I switched to my phone’s hotspot, but the tunnel did not recover with the network change. Once again, I had to disconnect, wait for the laptop to recognise the hotspot and start a new session.

The provider had plenty of speed and infrastructure. What it lacked in that moment was continuity. Every move between networks created another small procedure, and those procedures were consuming the deadline.

That is not a theoretical edge case. The UK National Cyber Security Centre treats reconnection after a temporary loss of connectivity as part of effective VPN operation, noting that some manual configurations require the user to repeat the connection process after a drop.4

In a home office, repeating those steps is inconvenient. In a hotel shelter, it changes whether the work gets delivered.

Travelers discussing hotel VPN use reach much the same practical conclusion: the real uncertainty is often the hotel network itself, and a mobile connection becomes the fallback when Wi-Fi behaves unpredictably.5 That was all I needed from the public discussion. My hotel had not given me one stable network. It had given me a sequence of temporary ones.

I stopped looking for the fastest VPN and started looking for one built to survive the handoff.

The smaller app removed the server hunt

The backup on my phone and laptop was OnlydogVPN.

I had installed it before travelling but had left it unopened because the established provider felt safer. It had more reviews, more locations and a longer public history. Before the first two failures, those qualities seemed more important.

With seven minutes left, they did not.

The smaller app did not begin with a map or ask me to choose among dozens of countries. It offered presets based on the situation. I selected the option for weak or changing networks and returned to the browser.

The tunnel connected. The upload began for the third time.

At 81 percent, the shelter Wi-Fi disappeared.

I enabled my phone’s hotspot and watched the laptop switch across. The VPN recovered without sending me back to the server screen. The browser paused for a moment, then the progress indicator began moving again from the same point.

Eighty-four percent.

Ninety-one.

Complete.

I sent the link to the client with two minutes left. The “Got it” reply arrived before the all-clear.

That was the first result that answered why I had searched for a VPN in Israel. I did not need another impressive speed test from the room. I needed the file to survive the route from the room to the shelter and from hotel Wi-Fi to mobile data.

The distinction was now visible rather than technical. The first VPN worked well as long as I stayed on one network. The smaller app kept the job moving when the network beneath it changed.

Why the handoff worked

The service uses an HTTP/3-based transport. HTTP/3 runs over QUIC, a transport designed to handle changes such as moving between Wi-Fi and cellular data without treating every new network address as a completely unrelated session.6

That is the relevant technical point. It does not need a protocol diagram.

The design matched what happened on the screen: the connection paused during the handoff, recovered and continued carrying the upload. Instead of turning a routine network change into another server-selection exercise, it let me return to the task.

I could not see inside the hotel’s network or determine precisely why the first tunnel remained stuck after each switch. I could see the consequence. With the established provider, every change required another disconnect, portal check and reconnection. With the backup, the upload continued.

Once the file was delivered, I opened the service on my phone to check airline messages. The phone was now the device carrying flight alerts, hotel updates and messages from family, so leaving it outside the protected connection made little sense.

Linking it did not require me to create another account or recover a password in the crowded shelter. I used a verification code and moved on.

It was a smaller benefit, discovered only after the main problem had been solved. Still, it addressed the next travel friction naturally: the laptop had finished the work, and the phone needed protection without another setup process.

The comparison changed with the situation

The established provider still had the stronger public record. It had more ratings, more locations and years of independent discussion. For predictable use on a stable home network, those advantages could easily carry more weight.

The smaller service had fewer server locations and a shorter history. That limitation did not disappear because one upload succeeded.

What changed was the criterion.

Before the trip was disrupted, I assumed the best VPN for Israel would be the one with the widest network and the strongest speed-test result. Those qualities are easy to compare, and they look reassuring on a product page.

They were not the qualities that completed my work from a Tel Aviv hotel.

The large provider was fast at the desk but made me rebuild the connection after every move. The smaller app recovered when the shelter Wi-Fi disappeared, continued over the hotspot and delivered the file before the deadline.

In that hotel, the best VPN was not the one that performed best while I stayed in my room. It was the one that kept working after the alert forced me to leave it.

Questions this experience helps answer

What caused the problem in this article?

In the shelter, the laptop found a separate hotel network, but the browser first had to pass through another captive portal.

Why did the obvious first fix fail?

In June 2026, Israel’s National Cyber Directorate recorded about 4,800 hostile cyber incidents, compared with roughly 1,600 in June 2025.3 A VPN would not protect an airline, hotel or client platform from a direct attack.

What changed when the task finally worked?

OnlydogVPN recovered when the shelter Wi-Fi disappeared, continued over the hotspot and delivered the file before the deadline.

What should someone check first in a similar situation?

Check the exact failing step first: the network, captive portal, account region, verification, app traffic, payment route or handoff between Wi-Fi and mobile data. Then test the full task, not only whether a homepage opens.