For most end users, connectivity is something they simply take for granted. Yet behind every stable connection and every device lies months of preparation. Recently, we completed one of our larger maintenance projects: upgrading our Belnet Service Routers (BSRs) and backbone routers — together forming the backbone of our network — around 360 devices spread across Belgium.
Why upgrade proactively?
Network equipment runs on software that requires regular updates. These updates bring security patches, bug fixes, improvements, and new features.
Regular updates are also essential to remain supported by the vendor. Over time, older software versions fall out of support, leaving you on your own — no fixes and no assistance when something goes wrong.
That’s why we chose to upgrade well in advance. Not because we were already facing issues, but to prevent them. When you are responsible for the connectivity of universities, research institutions, and hospitals operating 24/7, waiting is simply not an option.
Test first, then deploy
Before touching a single production device, we go through a thorough lab testing phase. For each platform — all four BSR models as well as the backbone routers — the upgrade is first performed on a test device.
We replicate the production environment (software version and configuration) and execute the upgrade following a clearly defined procedure. Before and after the upgrade, we collect and compare all relevant system outputs. Where needed, we refine the process.
Only once these tests are successful (which they were) do we proceed with the rollout to production devices. This lab phase is far from a formality: it gives us confidence and clearly defines the exact steps to follow for each platform.
What happens beforehand?
BSRs are the routers that directly connect our customers to the Belnet network — hence the name: Belnet Service Router. In total, more than 360 devices across four models and locations throughout Belgium. In parallel, we also upgraded our backbone routers, which handle large data flows between our core datacenters.
For each device, we first establish a full baseline: active connections, configuration, and normal operating metrics. This snapshot serves as our reference point afterward.
During testing, we sometimes find that configurations that worked perfectly on the old software are no longer compatible with the new version. We adjust them beforehand to avoid surprises during the actual upgrade.
Interventions are deliberately scheduled in the evening, between 8 PM and 1 AM, outside office hours. Even though the impact is minimal, we aim to minimize disruption for our users. This means our team regularly works late, while most of the world is offline.
During these interventions, engineers work directly on the devices while other colleagues monitor the network and keep a close eye on alerts. This coordinated approach ensures everything remains under control.
Not everything goes perfectly
Even with careful preparation, unexpected issues arise. For example, one of the candidate software versions showed a memory issue when combined with a specific BSR model.
Because we detected this early, we were able to quickly engage with the vendor and obtain a corrected version. On another device, we observed brief instability after the upgrade. There was no impact on customers, but it was enough to further refine our procedure.
That’s how we work: document, learn, and continuously improve.
The result
By the end of April 2026, all BSRs and backbone routers are running a recent software version that will remain supported for a long time. The network is more stable, more secure, and ready for the future.
Our users noticed little to nothing — exactly as intended. For us, as Belnet network engineers, that’s what it’s all about: a network that runs reliably day after day, without anyone having to think about it.
