The Campus VPN Worked in the Library. My Assignment Failed in the Dorm

At 11:18 p.m. in Shanghai, my group presentation was finished except for one thing: submitting it. The video sat in Google Drive at 63 percent, my professor’s final comments were in Gmail, and the other three students were waiting in WhatsApp for the upload link. I blamed the residence-hall Wi-Fi, moved my laptop closer to the door and restarted the transfer. It returned to zero.

The deadline was midnight.

I reopened the university VPN. Its icon turned green. Gmail displayed two new messages, then stopped loading. The Drive upload reached four percent and froze.

That morning, the same laptop had worked perfectly in the library.

The assignment had not changed. The network underneath it had.

The library and the dorm offered different versions of campus life

China’s universities were attracting international students again. More than 380,000 students from 191 countries and regions studied in China during the 2024–25 academic year, a 15 percent increase from the previous year. China’s next five-year education plans also placed renewed emphasis on expanding the “Study in China” programme.

Many of those students arrived with an academic life already built around services from home.

My course documents were in Google Drive. My university email used Gmail. Our group preferred WhatsApp because two members were European exchange students and another would return to the United States after the semester. The presentation software, reading list and meeting links had all been chosen before any of us reached China.

On campus, that arrangement often felt normal. Some universities provide managed international connectivity and institutional VPN access. NYU Shanghai, for example, warns students that Gmail, YouTube and other global services are inaccessible on ordinary Chinese connections while explaining the international access available through its own network. Yale offers similar guidance for students and faculty working in China.

That was why the morning had been so easy.

In the library, I had been using a managed university network.

In the residence hall, I was back on an ordinary local connection that interrupted the services my assignment depended on.

Once I understood that difference, reopening the official university VPN seemed like the responsible next step.

The university VPN restored access, then lost the upload

The institutional connection had clear advantages.

It came from the university responsible for my account. It gave me access to library databases and internal systems, and there was an IT department I could contact if something failed.

I entered my student ID, password and authentication code.

Gmail opened.

My professor’s message contained only two changes: correct a date on slide 14 and replace the final chart with the version attached to the email. I downloaded the chart, edited the slide and restarted the video upload.

Nine percent.

Seventeen.

Then the residence network asked me to sign in again.

The VPN dropped. Drive returned to zero.

I completed the Wi-Fi login page, reopened the university client and approved another authentication request. The upload began for the third time.

At 11:29, it stalled at 21 percent.

The official VPN had solved the first problem: I could reach the services. It had not solved the problem that now mattered more. Every residence-network interruption forced the connection to begin again, taking the upload with it.

That changed my standard.

I no longer cared which VPN produced the highest speed for two uninterrupted minutes. I needed one connection that would remain useful long enough to finish the assignment.

The large provider gave me more routes to test

I had also subscribed to a major commercial VPN before leaving home.

It had a long public history, a large support operation and servers across dozens of countries. Those were genuine strengths for a student spending a semester moving between airports, hotels, classrooms and rented rooms.

I opened the app and accepted its recommended nearby connection.

Gmail refreshed. Google Drive displayed the presentation folder, and the upload began again.

At first, it was faster than the university VPN. The progress indicator reached 28 percent in less than three minutes.

Then it stopped moving.

I selected another country. Drive loaded, but the upload failed before it began. A third route opened WhatsApp Web while leaving the file transfer spinning.

I could not observe the filtering rules inside the network, so I could not identify the precise reason each route behaved differently. What I could see was that the server list was giving me more attempts, not a more durable session.

China’s filtering does more than block a fixed collection of websites. Researchers have documented systems that inspect connection details and interfere selectively with encrypted traffic. The practical consequence was visible on my screen: a VPN could report that it was connected while the work behind it remained stalled.

At 11:38, I stopped changing countries.

I had 22 minutes left. I did not need a longer list of possible routes.

I needed one route that could carry the file to completion.

The backup moved the assignment instead of the map

I had installed OnlydogVPN^↗ before departure as a backup and then largely forgotten about it.

My hesitation had been reasonable. The service had fewer locations than the established provider, a shorter public history and fewer independent reviews. Someone who regularly needs a large catalogue of exact country endpoints would have more options with the bigger company.

That night, geography was no longer the decision I needed to make.

The smaller app organised its connections around situations rather than beginning with a world map. I selected the preset for a restrictive network and reopened Drive.

The upload started at zero for the final time.

Five percent became 16. Then 31.

WhatsApp refreshed beside it. One group member asked whether we should submit the smaller, lower-quality export instead.

I typed: “Give me five minutes.”

The upload passed 63 percent, the point where the first attempt had failed.

It reached 81, then 94.

At 11:47, Drive finished processing the video. I copied the link into the university submission page, checked the permissions and pressed Submit.

A confirmation number appeared.

I sent a screenshot to the group. It received three immediate reactions and one complaint that I had caused everyone unnecessary stress.

That was fair.

Only after the file was submitted did I look at what had changed. The smaller app uses an HTTP/3-based transport with additional obfuscation, helping the connection cross a restrictive network without presenting the same familiar VPN pattern that had repeatedly stalled.

The technical explanation did not need to be longer than the result.

The earlier connections had opened Gmail or Drive for a moment.

This one moved the video from my laptop to the professor’s submission page.

The dorm Wi-Fi failed again

Submitting the assignment should have ended the night.

Then one group member noticed that the opening video had no sound in the browser preview.

We started a quick call to check whether the actual file was damaged. The video played correctly on my laptop, and I began sharing my screen.

Two minutes later, the residence Wi-Fi disconnected again.

The call froze on a particularly unflattering frame of my face.

Earlier, that failure would have meant reopening the VPN, selecting another route and hoping the call returned before everyone gave up.

This time, I enabled my phone’s hotspot and moved the laptop onto mobile data.

The call paused.

Then the voices returned.

The video resumed from the same section, and my group members were still discussing the sound problem when the shared screen came back.

The app’s HTTP/3-based transport is designed to recover more smoothly when the network path changes. In practice, the dorm Wi-Fi failed, the laptop moved to the hotspot and the group call continued without another server search.

That was a smaller result than submitting the assignment, but it gave me a reason to keep the app installed afterward.

The first connection had completed the urgent task.

The recovery meant I would not have to solve the same problem every time my study location changed.

A study-abroad VPN has to follow the student

Before that night, I had treated a VPN as another item on a departure checklist.

Passport.

Power adapter.

Local SIM.

VPN.

That list made the choice sound permanent, as though installing one app before the flight would settle the internet question for an entire semester.

Study abroad does not happen on one network.

A student moves from a managed library connection to a residence hall, from a café to a train and from Wi-Fi to mobile data. University email, cloud documents and group chats may cross several of those networks in one afternoon.

Student discussions reflect the same practical divide: campus access works in one location, then research, writing or communication becomes another troubleshooting exercise elsewhere. The useful lesson is not that every student needs the same provider. It is that a VPN tested only on the easiest campus connection has not yet faced the conditions that usually cause the problem.

The university VPN remained the right tool for internal systems and protected academic resources.

The established commercial provider offered more locations, a longer history and broader public scrutiny.

The smaller app offered less visible scale. What it gave me was continuity: a route that completed the upload on the residence network and recovered when the laptop moved to mobile data.

That night, the useful study-abroad VPN was not the one that looked fastest in the library.

It was the one that stayed with the assignment after I left it.