Course contentsShow
← System Design
Lesson 506 of 1,91914. CAP TheoremFree lesson

CAP in Normal Operation vs Partition

During normal operation (no partition), systems can provide both consistency and availability; CAP trade-off only applies during partition events.

CAP in Normal Operation vs Partition

You'll learn: Why CAP theorem's trade-offs only matter when network partitions actually occur, not during normal system operation.

The Normal Operation Illusion

Here's a crucial insight many engineers miss: CAP theorem doesn't force you to choose between consistency and availability all the time. During normal operation—when all nodes can communicate freely—distributed systems can provide both strong consistency AND high availability simultaneously.

The CAP trade-off only becomes real when a network partition occurs: when nodes become isolated from each other and cannot communicate. Only then must you choose whether to:

  • Sacrifice availability (reject requests until partition heals) to maintain consistency (CP choice)
  • Sacrifice consistency (allow divergent data) to remain available (AP choice)

Think of It Like a Bridge

Imagine a city with multiple neighborhoods connected by bridges. During normal operation, you can travel anywhere reliably and get current information from any neighborhood (consistency + availability).

But when a bridge collapses (partition), you must decide:

  • Stop serving requests from isolated neighborhoods until the bridge is repaired (CP)
  • Let isolated neighborhoods operate independently, even if information becomes inconsistent (AP)

The Key Distinction

Most of the time, distributed systems run without partitions. Network failures are relatively rare. During these periods, properly designed systems deliver both consistency and availability without compromise.

CAP theorem tells you: plan for partition events, but don't assume you must sacrifice consistency or availability during everyday operation.

Key Takeaway: CAP is about handling exceptional partition scenarios, not normal operation—well-designed distributed systems achieve both consistency and availability when the network is healthy.