Course contentsShow
← System Design
Lesson 511 of 1,91914. CAP TheoremFree lesson

Banking Systems: Consistency Over Availability

Financial transactions typically choose consistency (CP), blocking writes during partitions to prevent data anomalies and double-spends.

Banking Systems: Consistency Over Availability

What You'll Learn

Why financial transaction systems deliberately sacrifice availability during network partitions to maintain data integrity.

The CP Choice in Banking

When a bank processes your transfer of $500 from checking to savings, consistency is non-negotiable. The system must guarantee that:

  • Money leaves one account
  • Money arrives in the other
  • No duplicate charges occur
  • Balances remain accurate

Banking systems are classic CP systems (Consistency + Partition Tolerance). During a network partition, they choose to block operations rather than risk incorrect account balances or double-spending.

Why Availability Takes a Back Seat

Imagine two datacenter halves lose communication. If both continued processing withdrawals independently:

  • Account balance: $1,000
  • Partition A processes: withdraw $800
  • Partition B processes: withdraw $800
  • When healed: which withdrawal wins? Both can't succeed.

Banks prevent this by rejecting transactions during partitions until the system can verify all data is consistent. You might see "Service temporarily unavailable" — frustrating, but far better than lost money or phantom balances.

The Trade-off in Action

During normal operation, banking systems run smoothly. But when partitions occur:

  • Writes are blocked to prevent conflicting updates
  • Reads may also block to prevent stale balance information
  • Customers experience downtime
  • Financial accuracy is preserved

This is the opposite of social media feeds (AP systems), which stay available by showing possibly stale data. Banks can't show "stale" account balances — every cent must be accounted for correctly.

Key Takeaway: Banking systems prioritize consistency over availability because financial correctness is more critical than continuous uptime; temporary service disruption is acceptable, but incorrect balances are catastrophic.