Interview Question: Explain CAP Theorem

1. What is the CAP Theorem?

The CAP Theorem states that a distributed system can only guarantee two out of three properties at the same time:

1. Consistency (C): All users see the same data simultaneously.
2. Availability (A): The system always responds to requests (even with stale data).
3. Partition Tolerance (P): The system works even if parts of it lose communication.

Key Insight: Network partitions (e.g., server crashes, dropped messages) are unavoidable. During a partition, you must choose between C and A.

2. Real-World Example: Rohit’s “Reminder” Startup

Imagine Rohit runs a company called Reminder where he writes down customer reminders in a diary. When he hires Raj to help, problems arise:

Problem 1: Inconsistency

• Scenario: Rohit writes a reminder in his diary, but Raj doesn’t update his copy.
• Result: Customers get inconsistent information.

  [Client] --> [Rohit's Diary: "Call X"]  
  [Client] --> [Raj's Diary: "No Entry"]  
  

• Solution: Both must write entries simultaneously to ensure consistency.

Problem 2: Availability Issues

• Scenario: If Raj is unavailable, Rohit can’t respond to requests until Raj returns.
• Result: Delays frustrate customers.

  [Client Request] --> [Rohit: Available ✅]  
                     [Raj: Unavailable ❌]  
  

• Solution: Accept entries even if one person is missing, and sync later.

Problem 3: Network Partition

• Scenario: Rohit and Raj can’t communicate. The system must either:
  • Reject requests (prioritize consistency but lose availability).
  • Accept requests (prioritize availability but risk inconsistency).

  Partition Occurs:  
  [Rohit] --✂️ Network Failure ✂️-- [Raj]  
  Choose:  
  1. Reject writes ➔ Consistency ✔️, Availability ❌  
  2. Accept writes ➔ Availability ✔️, Consistency ❌  
  

3. CAP Trade-offs: CP vs. AP vs. CA

⚠️ CA systems are theoretical – real-world distributed systems must handle partitions (P).

4. PACELC Theorem: Beyond CAP

The PACELC Theorem extends CAP to address normal operations (no partitions):

• PAC: During a partition (P), choose between A (availability) and C (consistency).
• ELC: Else (E), choose between L (latency) and C (consistency).

Examples

• Banking Apps: Prioritize consistency (e.g., ATMs must show accurate balances).
• Social Media: Prioritize availability (e.g., Facebook News Feed can show stale posts).
• Messaging Apps: Prioritize consistency (e.g., WhatsApp must show messages in order).

5. CAP in Popular Databases

Key Takeaways

✅ CAP Theorem: Choose 2/3 properties during a partition.
✅ PACELC: Adds latency vs. consistency trade-offs in normal conditions.
✅ AP vs. CP: Match the system to the use case (e.g., social media = AP, banking = CP).

Further Reading:

• CAP Theorem Explained (BMC)
• PACELC Deep Dive (Educative)