Cloud Native Go, 2nd Edition

Chapter 4. Cloud Native Patterns

Progress is possible only if we train ourselves to think about programs without thinking of them as pieces of executable code.¹

Edsger W. Dijkstra, August 1979

In 1991, while still at Sun Microsystems, L Peter Deutsch² formulated the “fallacies of distributed computing,” which lists some of the false assumptions that programmers new (and not so new) to distributed applications often make:

• The network is reliable: switches fail, routers get misconfigured

• Latency is zero: it takes time to move data across a network

• Bandwidth is infinite: a network can handle only so much data at a time

• The network is secure: don’t share secrets in plain text; encrypt everything

• Topology doesn’t change: servers and services come and go

• There is one administrator: multiple admins lead to heterogeneous solutions

• Transport cost is zero: moving data around costs time and money

• The network is homogeneous: every network is (sometimes very) different

If I might be so audacious, I’d like to add a ninth one:

• Services are reliable: services that you depend on can fail at any time

In this chapter, I’ll present a selection of idiomatic patterns—tested, proven development paradigms—designed to address one or more of the conditions described in Deutsch’s fallacies and demonstrate how to implement them in Go. None of the patterns discussed in this book are original to this book—some have been around for as long as distributed applications have existed—but ...