Course contentsShow
← Software Engineering
Lesson 1111 of 1,93333. Error Handling and ResilienceFree lesson

Introduction to the Fallacies of Distributed Computing

Understanding the eight false assumptions developers commonly make about distributed systems and their impact.

Introduction to the Fallacies of Distributed Computing

You'll learn the eight dangerous assumptions developers commonly make when building distributed systems—and why these assumptions lead to fragile, unreliable software.

The Eight False Assumptions

When Peter Deutsch and colleagues at Sun Microsystems first articulated the "Fallacies of Distributed Computing" in the 1990s, they identified eight assumptions that seem reasonable but are fundamentally wrong:

  1. The network is reliable — Networks drop packets, cables break, and connections fail constantly.
  2. Latency is zero — Every network call takes measurable time; remote isn't the same as local.
  3. Bandwidth is infinite — Network capacity is limited and often shared with competing traffic.
  4. The network is secure — Traffic can be intercepted, modified, or redirected.
  5. Topology doesn't change — Servers move, get replaced, and network paths shift regularly.
  6. There is one administrator — Multiple teams, vendors, and organizations control different parts.
  7. Transport cost is zero — Serialization, network overhead, and infrastructure cost money and time.
  8. The network is homogeneous — Different protocols, platforms, and technologies must interoperate.

Why This Matters

Think of these fallacies like assuming all roads are perfectly paved, traffic-free, and safe. Your car (application) might work fine in that fantasy world, but real driving requires handling potholes, congestion, and accidents. Similarly, distributed systems must actively handle network realities, not ignore them.

These fallacies directly relate to the resilience patterns you've learned. Timeouts address latency and reliability. Circuit breakers prevent cascading failures when dependencies fail. Bulkheads isolate failures across topology changes. Each pattern exists because one or more of these fallacies would otherwise cause system failures.

Key Takeaway: The eight fallacies of distributed computing are false assumptions about network behavior that, when believed, lead to brittle systems; recognizing them is the first step toward building truly resilient distributed applications.