The next network bottleneck is hiding in the uplink Tobie van Schalkwyk November 2, 2026

The next network bottleneck is hiding in the uplink

6

By Tobie van Schalkwyk, BU Executive at Duxbury Networking

It is easy to focus on the access layer when upgrading a network. Faster access points and devices naturally lead to faster switch ports. But all of that traffic still has to move upstream, and an older uplink can quickly become the part of the network holding everything back.

The Ethernet Alliance’s 2026 Ethernet Roadmap points to growing adoption of 2.5G, 5G, and 10GBASE-T in enterprise networks, alongside faster optical uplinks. Wi-Fi 7 is part of that, with more access points capable of generating enough traffic to justify multi-gigabit wired connections.

For resellers, the real design question is whether uplink capacity has kept pace with faster access ports and the traffic they now carry.

Faster access changes the uplink calculation

Consider a 24-port switch in which every access port can operate at 2.5Gbps. That represents 60Gbps of potential access capacity. If the switch has a single 10Gbps uplink carrying traffic towards the rest of the network, the theoretical oversubscription ratio is 6:1.

That does not automatically make the design unsuitable. Very few networks have every connected device transmitting at full port speed simultaneously, and oversubscription has always been part of network design.

What counts is how the network is actually being used. A network with mostly light user activity will place very different demands on the uplink from one carrying busy wireless, surveillance, or other data-heavy workloads.

Oversubscription needs context

Oversubscription is normal. A better guide is what the uplink is actually carrying when the network is busy, rather than the combined maximum speed of every access port. It is also worth understanding how traffic moves through the environment. Some communication may remain local to the switch or access layer, while traffic heading towards servers, internet services, other network segments, or centralised resources has to use the uplink.

The design should leave enough headroom for busy periods and realistic growth without specifying capacity the customer is unlikely to use.

This is especially relevant when newer access switches are introduced into an existing network. Increasing edge capacity without reviewing the path upstream can simply move the constraint elsewhere.

Check the fibre path before increasing uplink speed

Moving from 10Gbps uplinks to 25Gbps or faster may appear straightforward when the switches support it, but the physical path also needs attention.

The transceivers at each end have to be compatible with the equipment and the fibre between them. Fibre type and link distance can determine which speeds and optics are practical, particularly where older cabling is already installed. A switch upgrade can therefore expose limitations in the existing fibre path that were not obvious before.

Plan for the network under pressure

A design using redundant uplinks may have comfortable capacity while both links are available. If one fails, the remaining path has to carry more of the load. Cisco’s campus design guidance recommends assessing oversubscription during failure conditions as well as normal operation, because available uplink bandwidth can change considerably when part of the design is lost.

Existing utilisation data is a useful starting point, but the design also has to allow for how the network is likely to change over time.

Look beyond the access port

The Ethernet Alliance’s recent analysis of 2.5GBASE-T and 5GBASE-T adoption notes that 1Gbps still dominates enterprise campus ports, while 2.5Gbps and 5Gbps show the strongest forecast growth. For resellers, that shift changes the arithmetic behind the network and makes uplink planning harder to ignore.

DuxNet works with partners on switch selection, uplink planning, optics, and network design so that improvements at the access layer are supported by the infrastructure behind them.

A faster access port delivers its value only when the rest of the network can carry the traffic it creates. As multi-gigabit access becomes more common, the uplink deserves the same scrutiny as the ports facing the user.

Scroll to Top