Course contentsShow
← System Design
Lesson 518 of 1,91914. CAP TheoremFree lesson

PC/EC Systems: Consistency Always

Systems like traditional RDBMS and HBase that prioritize consistency even at the cost of latency during normal operation.

PC/EC Systems: Consistency Always

What You'll Learn

You'll understand systems that choose consistency over both availability (during partitions) and low latency (during normal operation).

The Double Commitment to Consistency

In the PACELC framework, a PC/EC system makes consistency its top priority in both scenarios:

  • PC: During a partition, it sacrifices availability to maintain consistency
  • EC: Even during normal operation (the "Else" clause), it sacrifices latency for consistency

These systems will wait, block, or reject requests rather than serve stale or inconsistent data—ever.

The Core Philosophy

Think of a bank vault with multiple security checkpoints. Even when everything's working fine, you must wait through each verification step. During an emergency lockdown (partition), access is completely denied rather than risk a security breach.

PC/EC systems operate similarly: consistency is non-negotiable.

Classic Examples

Traditional RDBMS (PostgreSQL, MySQL with strong settings):

  • Always enforce ACID guarantees
  • Queries wait for locks to resolve
  • Replication lag blocks reads if configured strictly
  • Higher latency than eventual consistency systems

HBase (with strong consistency settings):

  • Single Region Server ownership per row ensures consistency
  • Writes wait for WAL sync before acknowledging
  • Reads always hit the authoritative Region Server
  • No stale reads, but higher latency costs

When to Choose PC/EC

Use PC/EC when:

  • Correctness trumps speed: financial transactions, inventory counts
  • Stale data is unacceptable: medical records, legal documents
  • Auditing requires precision: regulatory compliance systems

The trade-off: every operation pays a latency tax for guaranteed correctness.

Key Takeaway: PC/EC systems choose consistency in all circumstances—sacrificing availability during partitions and accepting higher latency during normal operation to ensure data is always correct and up-to-date.