I’ve noticed a recurring pattern with networking-heavy applications: everything works as expected on a local machine or VM, but the moment it’s deployed inside Docker or Kubernetes, service discovery starts failing.

A recent example involved a BACnet/IP application that could discover devices normally outside Kubernetes, but failed to find anything once it was running inside a pod.

It made me wonder whether we’re hitting a broader mismatch between modern cloud-native infrastructure and protocols that were originally designed for local networks.

Some questions that come to mind:

• Are multicast and broadcast-based discovery protocols fundamentally at odds with Kubernetes networking?
• How do teams typically handle legacy industrial, IoT, or building automation protocols in containerized environments?
• Do you adapt the network architecture, or is it sometimes better not to containerize these workloads at all?
• What networking assumptions do older protocols make that cloud-native platforms often break?

Interested in hearing from anyone who’s had to bridge the gap between traditional network protocols and modern orchestration platforms. What was the biggest lesson you learned?