Single-Hop vs Multi-Hop VPN: The Extra Relay Protected More Than My Call Could Survive

The source’s face appeared for half a second.

Then the call froze.

I was sitting in an empty meeting room beneath a technology conference in Berlin. Upstairs, several hundred people were sharing the same Wi-Fi. Downstairs, I had twenty-five minutes to interview someone who had agreed to speak only during a narrow break in their workday.

The conversation was sensitive enough that I had enabled my established VPN provider’s multi-hop mode. My connection entered through Sweden and exited through Switzerland before reaching the encrypted call room.

The setup looked reassuring.

The call did not.

My source’s first sentence arrived in pieces. Their mouth moved, the audio followed two seconds later, and the browser displayed:

Your connection is unstable.

I blamed the conference network.

I turned off the camera, closed every unnecessary tab and restarted the call.

The source rejoined.

This time, I heard three complete words before the audio stopped.

My speed test showed 86 Mbps.

The interview remained impossible.

Multi-hop solved a real privacy problem

A normal single-hop VPN sends traffic through one VPN server before it reaches its destination.

Multi-hop adds another relay. The first server receives the connection from the user, while the second becomes the exit seen by the website or service.

That separation can be valuable when someone is concerned about a hostile network, a compromised server or attempts to compare traffic entering and leaving the VPN infrastructure. Major providers offer multi-hop modes specifically for those higher-risk situations. (Protonvpn)

The idea made sense for the interview.

The conference network could see that my laptop was using an encrypted connection, but it could not read the call. The call platform saw the second VPN server rather than the conference Wi-Fi.

The second relay created more distance between those two observations.

It also created more distance between every question and answer.

My traffic was travelling from Berlin to Sweden, then Switzerland, then the call service and back again. That longer route added delay to a conversation that depended on voices arriving in real time.

The privacy benefit was real.

So was the silence.

The fastest pair still arrived too late

I changed the multi-hop combination.

The provider offered enough entry and exit locations to make optimisation feel possible. I tried a German entry with a Swiss exit.

The call opened.

The delay was long enough for both of us to begin speaking at once, stop, apologise and repeat the same mistake.

I tried Germany to the Netherlands.

The source’s voice became clearer, but screen sharing never finished loading.

A nearby pair produced the best speed-test result of the morning. It also introduced brief losses that made every longer answer sound as though someone had removed random words.

Other users have described the same practical limit: ordinary browsing remains comfortable through a multi-hop connection, while real-time calls become difficult to follow. (Reddit)

That was enough to confirm what the interview had already shown.

The multi-hop connection was active.

It was simply too delayed to carry a natural conversation.

I had started with the assumption that more routing meant more protection.

Ten minutes into the interview window, I realised protection was useful only if the source could finish a sentence.

Single-hop fixed the delay but not the session

I disabled multi-hop and selected the established provider’s recommended single server.

The difference was immediate.

The call opened quickly.

The source’s voice matched their face.

We completed the first question without speaking over each other.

For several minutes, single-hop looked like the correct answer. The traffic still travelled through a private tunnel, but it avoided the extra relay that had made the call unusable.

Then the conference Wi-Fi dropped briefly.

The laptop reconnected to another access point with the same network name. Normal webpages recovered, but the VPN remained on Reconnecting.

The call room ended the session.

When I reopened it, the source was waiting.

I reconnected the VPN and tried again.

The browser returned an error before the call loaded.

A second single-hop server opened the page but lost the audio stream.

A third showed as connected inside the VPN app while the call room remained blank.

The problem had changed.

Multi-hop had been too slow.

The single-hop route was fast enough, but it could not recover cleanly when the crowded conference network shifted underneath it.

I now had nine minutes left.

The comparison was no longer about which diagram looked more private.

It was about which connection could keep the source in the room.

Mobile data moved the problem into the corridor

I disconnected from conference Wi-Fi and turned on my phone hotspot.

The established single-hop connection opened over mobile data.

The source returned.

For the first minute, the call was clear.

Then the phone dropped from 5G to a weak indoor signal. The browser lowered the audio quality, paused and finally removed me from the room.

I moved toward the corridor, where reception was slightly better.

The conference noise upstairs made a sensitive interview impossible.

The hotspot avoided the unstable public network, but one bar of mobile service could not carry the conversation from a concrete basement.

I could use the quiet room with unreliable Wi-Fi or the noisy corridor with unreliable cellular data.

Neither completed the interview.

At that point, the extra separation offered by multi-hop had become irrelevant. It could not compensate for a route that prevented the source from speaking.

I needed one private connection the conference network would carry—and one that recovered before the call platform gave up.

The source completed the answer through the smaller route

I closed the established provider and opened OnlydogVPN.

The smaller app did not ask me to choose an entry country, an exit country and a server pair. I selected the situation for a private call on a restrictive public network and connected.

Then I reopened the interview room.

The source appeared.

The audio followed immediately.

I asked the question that had been interrupted three times.

They answered for nearly four minutes.

No words disappeared.

The browser did not lower the call quality or display another network warning.

We continued through the remaining questions. At one point, the source shared a document on screen. The text sharpened, stayed readable and advanced as they moved between pages.

The call timer passed ten minutes.

Then fifteen.

By the time we finished, I had stopped watching the connection indicator.

The interview had become an interview again.

Only after the source completed the final answer did the route design matter. The smaller service uses an HTTP/3-based transport with additional traffic obfuscation, giving the call a responsive route through the busy conference connection without adding a second relay.

I could not inspect the conference network’s internal filtering rules or identify the exact cause of every earlier failure. The visible result was decisive: multi-hop added too much delay, the established single-hop route broke when the network shifted, and the smaller app carried the complete conversation.

The document continued when Wi-Fi disappeared

After the source left the call, they sent a supporting document through the secure workspace.

I began downloading it before leaving the meeting room.

The conference Wi-Fi weakened as I climbed the stairs.

My laptop moved onto the phone hotspot.

The download paused briefly.

Then it continued from the same point.

The VPN remained active without asking me to select another server or restart the session. Its HTTP/3-based connection recovered as the underlying network changed. (IETF)

The result was simple:

Conference Wi-Fi disappeared.

Mobile data took over.

The document kept moving.

That recovery solved a smaller problem than the interview, but it arrived naturally after the main task had succeeded.

The source had already spoken.

Now the supporting material reached my laptop without another login, another server pair or another direct connection.

Multi-hop still belonged to a narrower threat

The failed call did not make multi-hop pointless.

There are situations where separating entry and exit traffic matters more than speed.

A journalist working under targeted surveillance may accept slower performance for stronger resistance to traffic comparison.

An activist connecting from a hostile network may deliberately choose one entry region and another exit region.

Someone who believes a particular VPN location could be monitored may value the additional separation provided by a second relay.

Those are focused reasons to use multi-hop.

They are not reasons to enable it automatically for every task.

A second relay adds another server and a longer path. For ordinary browsing, that cost may be easy to ignore. For a live interview, game, remote desktop session or urgent upload, it can become the reason the task fails.

The practical question is not simply whether multi-hop offers more separation.

It is whether that extra separation addresses the threat in front of me—and whether the work can survive the delay.

My interview could not.

Single-hop was faster, but speed alone was not enough

The established provider’s single-hop mode removed most of the delay.

That proved something useful.

For real-time work, one relay was better than two.

But the morning also showed why “choose single-hop” was not a complete answer.

The route still had to remain usable when the underlying network changed. When conference Wi-Fi moved the laptop between access points, the established tunnel did not recover before the call session ended.

The smaller app did not succeed by adding another privacy layer.

It succeeded by making one private route less fragile.

That was the criterion I had missed while comparing server diagrams.

A second relay can create more separation.

It cannot rescue a conversation from a connection that never becomes stable enough to carry it.

The useful route completed the conversation

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

Neither limitation affected the interview.

The established provider’s multi-hop mode offered meaningful additional separation, but the extra relay made the conversation too delayed to conduct.

Its single-hop mode restored natural audio, then lost the session when the conference network changed.

The smaller app asked fewer routing questions, opened the call, carried the source’s complete answers and kept the supporting document moving when Wi-Fi gave way to mobile data.

I entered the meeting room believing the safest connection was the one with the most relays.

I left with the interview because the better route was the one the conversation could survive.