Last March, I was lucky enough to attend an event hosted by Juniper Networks™ and Arrow ECS™ in Finland with a dozen of the most knowledgeable data network gurus. It was an incredible opportunity to exchange ideas and learn from one another. But what stood out was our follow-up discussion in a Finnish sauna. As we discussed the promises and fulfillments of 5G technology, the conversation started to heat up. We all agreed that the world needed a broadband comparison focused on user experience, explicitly analyzing applications and TCP latency with different broadband technologies. And now, thanks to our efforts, the real impact of 5G latency has finally been tested (Fact or fiction). I’m proud to have been a part of this groundbreaking work and can’t wait to see where it takes us next. 5G latency finally tested (Fact or fiction)

5G’s low latency was fact or fiction

Regarding 5G technology, low latency has been touted as a game-changer. But is it all it’s cracked up to be? Some early marketing messages claimed that 5G could provide a latency of just 1-2 milliseconds, but as the technology has evolved, it’s become clear that things aren’t that pretty simple. With 5G now Finland’s primary mobile service offering, it’s an excellent time to examine latency’s real impact and whether the marketing message around 5G’s low latency was fact or fiction.

Test environment

After the test, we determined the latency difference between the 4G, 5G, and fixed broadband connections. The new ZTE™ modem, which is available from every Finnish mobile operator, was used alongside a 4G (100Mbps) and 5G (100Mbps) mobile data connection (SIM), a fixed 100Mbps broadband connection (Huawei CPE), and a Windows 11 Pro workstation. To ensure accuracy, we disabled some resource-intensive functions on the workstation during the test, such as cloud synchronization. Additionally, we used a separate Network TAP device to copy the network traffic (packets) for analysis. The test results showed varying latency levels between the different connections, which will be valuable information for optimizing their network performance.

Phase 1 – Capacity Verification

During the tests, we measured the theoretical maximum speed of transmission and reception using the operator’s Speedtest software. The results were displayed as clear peaks in the graph. One important observation we made was that the capacity utilized by applications was significantly lower compared to the offered connection capacity. This suggests that there is room for improvement in optimizing the usage of available resources. 5G is 100Mbps, however, operator hasn’t limited its speed.

Picture 1 – Speedtest vs. application requirements

Phase 2 – Latency Results

When it comes to measuring the actual end-to-end delay and user experience of various application services, there are a lot of factors to take into consideration. One major factor is the distance between the workstation and the server, but other variables are also at play. For instance, many of the services we tested are distributed globally through cloud services, and third-party CDN services are often used during the initial handshake. User authentication and license verification may also occur in a different location than the application service provision. To get a comprehensive analysis, we measured all servers involved in the application service and calculated their averages.

Picture 2 – TCP server latency (4G, 5G, Cable Fixed)

Based on the findings, it is evident that a fixed broadband connection offers the least amount of delay. The latency of the 5G network is marginally better than that of the 4G network, although there are some instances where this is not the case. It’s fascinating to note that the maximum speed at which individuals can process incoming visual stimuli is roughly 13 ms.

How can latency be improved?

After conducting the initial phase, it has been determined that the delay is not due to congestion or capacity issues with the local operator’s network. Plenty of local connection capacity is available that exceeds the applications’ requirements.

However, it has been found that the DNS name server plays a significant role in latency and user experience. Ensuring that users are directed to the nearest cloud service is crucial. Additionally, the number and quality of hops in the operator’s network and their selected partners, the use of CDN services, the geographical location of data centers, and the network architecture of the service center and applications all affect latency.

One example of this is PayPal™ versus a local bank. Although PayPal™ services are located thousands of kilometers from Finland, their response time is still faster than local services for transactions, reports, products, and invoices.

It’s important to note that the user experience decreases as latency increases. This is evident in the table that shows the highest latencies per network technology during the test. Latency is influenced by many factors besides the broadband connection.

Picture 3 – Max application latency (4G, 5G, Cable Fixed)

After a thorough test, it has been concluded that the 5G connection shows slightly better latency than the 4G connection. However, the difference in user experience is hardly noticeable. It is suggested that the reason for sluggishness in applications should be looked for elsewhere rather than trying to improve the connection or increase its speed. The packets analyzed by the analyzer do not lie. While user experience can be subjective, it is essential to note that the marketing message claiming 1-2 ms latency was highly misleading.

In addition to this, we also analyzed the latency and quality of IoT, enterprise networks, NaaS, and SaaS services. If you’d like to learn more about our findings, please refer to the link provided.


Hannu Rokka

5Feet Networks Oy, Senior Advisor