This one became especially true in the microservices era. Autoscaling and the emergence of the cattle, not pets approach to managing infrastructure mean that the topology will constantly change. This can affect latency and bandwidth, so some of this fallacy's outcomes are the same as the ones described earlier.
Fortunately, the mentioned approach also comes with guidelines on how to effectively manage your herd of servers. Relying on hostnames and DNS instead of hardcoding IPs is a step in the right direction, and service discovery, described later in this book, is another one. A third, even bigger, step is to always assume your instances can fail and automate reacting to such scenarios. Netflix's Chaos Monkey tool can also help you test your preparedness.