Under the Hood: Network Upgrades in EU1.AMS and EU9.DUS

During the last two weeks, our two engineering teams completed major network upgrades in EU1.AMS and EU9.DUS.

That means our dual datacenter locations in Amsterdam and Meppel, Netherlands, plus Düsseldorf, Germany, received a healthy dose of new networking power, fresh switching capacity, and the kind of infrastructure work that usually involves careful planning, precise labeling, serious coffee consumption, and at least one cable that looks exactly like another cable but is absolutely not the same cable.

The good news is that, for most users, this maintenance was almost invisible. That was the plan, and we like when the plan survives contact with reality, because reality in datacenters has a rich sense of humor.

We replaced and added a lot of networking gear, expanded switch clusters, retired older equipment in the Netherlands, and completed the work without turning the maintenance window into a dramatic episode of “Where Did My Packets Go?”

So let’s look under the hood.

From One or Two Switches per Rack to Thousands of Ports per Location

More than a decade ago, things were simpler. In most locations, and there were not many of them back then, we typically used one or two switches per rack. Those were usually Juniper EX switches with gigabit ports and gigabit uplinks, and for that time they were more than enough.

Yes, gigabit was once “plenty.” Please do not laugh too loudly, some of those switches can still hear you.

We still use Juniper today, but of course the models are much newer, much faster, and much more powerful. Our infrastructure has grown a lot since those early days: we now operate across many more datacenter locations, serve far more customers, and manage thousands of active ports per location.

At that scale, dozens or hundreds of standalone switches are not “simple infrastructure” anymore. They are a small administrative kingdom with its own weather system. This is why we use Virtual Chassis technology in many locations.

In simple words, Virtual Chassis allows several physical switches to work as one logical device, managed from a single point. Instead of treating every switch as a separate island, we combine them into a single managed cluster, which makes provisioning, monitoring, configuration, and customer-specific changes much easier and much faster.

Why Virtual Chassis Makes Life Better

One of the best parts of this approach is how cleanly a switch cluster can be expanded. When more capacity is needed, we can connect an additional switch, integrate it into the cluster, and within a short time it becomes part of the same managed infrastructure.

The service management system sees the new ports, and they can be assigned to new servers without building a tiny network temple around every rack.

This matters a lot for real-world operations. When new dedicated servers are installed, when virtualization nodes are added, when private VLANs are requested, or when special customer configurations are needed, the work becomes much more predictable.

And yes, enabling a private VLAN for a customer no longer requires a heavy sigh from the network engineer followed by a sad walk to get coffee. It can be done quickly, cleanly, and without negotiating with ancient Layer 2 spirits.

Another useful feature is that Virtual Chassis setups can be hybrid. In several locations across Europe and the US, we mix switches with different configurations inside the same design. For example, racks with virtualization compute nodes and high-performance dedicated servers may use 10 Gbps and 40/100 Gbps switching, while simpler switches can still handle regular dedicated servers and service tasks.

This gives us flexibility without turning network management into a spreadsheet-powered horror movie.

What Changed in EU1.AMS and EU9.DUS

The recent upgrades in the Netherlands and Germany were similar in architecture, but different in what exactly had to be done.

In EU9.DUS, our team expanded the existing switch cluster by adding new networking devices. This gives the Düsseldorf location more capacity for new servers, more room for future growth, and better flexibility for customer configurations. The datacenter page includes location details, available services, and ordering links for customers who want to deploy there.

In EU1.AMS, which includes our dual datacenter presence in Amsterdam and Meppel, we went further and finally retired the previous generation of 1/10G switches. These devices served successfully and reliably for many years, and they deserve respect, but at some point even good hardware gets its well-earned retirement. The EU1.AMS page also includes location details, available services, and direct ordering options.

No dramatic farewell ceremony was held, although we admit that some old switches have more uptime history than many modern web apps have releases.

The result is a cleaner, more powerful, and more scalable network layer in both locations.

Why Most Users Did Not Notice Anything

Large network maintenance is not magic. It only looks quiet when a lot of work happens before anyone touches production gear.

Before the maintenance windows, our teams prepared the configuration, checked cabling plans, validated topology, scheduled work carefully, and made sure the changes could be introduced with minimal interruption. The goal was not only to upgrade the network, but to make the upgrade as boring as possible for customers.

In infrastructure, “boring” is a compliment.

A good maintenance is the one where engineers do serious work, monitoring stays calm, users continue doing their thing, and nobody needs to open an emergency ticket titled “Everything Is On Fire.” In this case, the upgrades were almost unnoticed by users, which is exactly how we wanted it.

More Capacity, Cleaner Management

These upgrades are part of the same long-term direction we follow across ITLDC: stronger network foundations, better scalability, faster provisioning, and infrastructure that can grow without becoming harder to operate.

For customers, this means more room for VDS, dedicated servers, private networking, custom setups, and future expansion in key European locations. For our engineers, it means a cleaner operational model and fewer reasons to stare at a rack wondering who invented unmanaged complexity and why.

As always, our in-house support team is online 24/7, our engineers are watching the dashboards, and the packets are moving.

Which, in networking, is basically poetry.