🔷 1. Redundancy (Core Idea)

🧠 Definition (Interview Opening)

"Redundancy is duplicating critical components to eliminate single points of failure and improve reliability."

---

⚡ Why it matters

• Prevent data loss
• Ensure system keeps running
• Improve availability

---

🎯 Analogy

💾 Google Docs:

• Your doc saved on multiple servers → one server fails → no data loss

---

🎤 FAANG Question

Q: Why is redundancy important? A:

"It removes single points of failure and ensures high availability during failures."

---

🚨 2. Single Point of Failure (SPOF)

👉 Anything whose failure = system down

Example

• One DB → crash = system down ❌
• Replicated DB → failover ✅

---

🎤 FAANG Question

Q: How do you remove SPOF? A:

"By introducing redundancy via replication, multiple instances, and failover mechanisms."

---

🔷 3. Replication (Core Concept)

🧠 Definition

"Replication is copying data from a primary node to one or more replicas to improve availability, fault tolerance, and scalability."

---

🔧 Basic Flow

Primary (write) → Replicas (read)

---

⚡ Why Replication?

• High availability
• Read scaling
• Disaster recovery

---

🎤 FAANG Question

Q: Why use replication? A:

"To improve availability, scale reads, and ensure durability of data."

---

⚙️ 4. Types of Replication (VERY IMPORTANT)

---

🟢 1. Synchronous Replication (Strong Consistency)

🧠 Idea

"Write succeeds only after replicas confirm"

---

✅ Pros

• Strong consistency
• No data loss

❌ Cons

• High latency
• Slower writes

---

🎯 Analogy

✍️ Writing + photocopy instantly before submitting

---

🎤 FAANG Question

Q: When use synchronous replication? A:

"When consistency is critical, like banking systems."

---

🔵 2. Asynchronous Replication (Eventual Consistency)

🧠 Idea

"Primary writes first, replicas update later"

---

✅ Pros

• Fast writes
• High availability

❌ Cons

• Data lag
• Possible data loss

---

🎯 Analogy

📩 WhatsApp message → delivered later

---

🎤 FAANG Question

Q: Risk of async replication? A:

"Replica lag can cause stale reads or data loss on failure."

---

🟡 3. Semi-Synchronous (Balanced 🔥)

🧠 Idea

"Wait for at least one replica, others async"

---

✅ Pros

• Better consistency than async
• Better performance than sync

---

🎤 FAANG Question

Q: Why semi-sync is used? A:

"To balance consistency and latency in production systems."

---

⚖️ 5. Comparison (Must Remember)

Type	Consistency	Latency	Data Loss
Sync	Strong	High	No
Async	Weak	Low	Possible
Semi-sync	Medium	Medium	Low

---

🔥 6. Missing but CRITICAL Concepts

---

🧠 A. Read vs Write Scaling

• Writes → Primary
• Reads → Replicas

👉 "Scale reads horizontally"

---

🧠 B. Replica Lag (VERY IMPORTANT)

👉 Delay between primary & replica

Problem:

• User sees stale data

---

🎤 FAANG Question

Q: How handle stale reads? A:

"Use read-after-write consistency or route critical reads to primary."

---

🧠 C. Failover (VERY IMPORTANT)

👉 When primary fails:

• Promote replica → new primary

---

Types:

• Manual
• Automatic (preferred)

---

🎤 FAANG Question

Q: What happens when primary fails? A:

"A replica is promoted to primary via failover mechanisms."

---

🧠 D. Multi-Leader / Leaderless (ADVANCED 🔥)

Single Leader (most common)

• One write node

Multi-leader

• Multiple writable nodes

Leaderless (e.g., DynamoDB)

• No primary

---

🎤 FAANG Question

Q: Why not always multi-leader? A:

"Because it introduces conflict resolution complexity."

---

🧠 E. CAP Theorem Connection 🔥

• Sync → CP (consistency)
• Async → AP (availability)

---

⚠️ 7. Trade-offs

Benefit	Cost
High availability	Complexity
Fault tolerance	Replication lag
Read scaling	Consistency issues

---

🚀 8. Real System Example (FAANG Thinking)

Example: Instagram

• Primary DB → writes
• Replicas → serve feed

👉 High read traffic → replicas handle load

---

🎤 9. FAANG Interview Script

Start

"To improve availability and scalability, I'll introduce database replication."

---

Explain

"Writes go to primary, and replicas serve read traffic."

---

Trade-off

"Async replication improves performance but introduces eventual consistency."

---

Add Depth

"We can use semi-sync to balance latency and consistency."

---

Failure Handling

"In case of failure, a replica will be promoted via failover."

---

Close Strong

"This removes single point of failure and enables horizontal scaling."

---

🧠 Final One-Line (Must Memorize)

"Replication improves availability and read scalability by duplicating data across nodes."

---

💡 FAANG-Level Insight (DIFFERENTIATOR)

"Replication is not just about backup — it's about scaling reads and surviving failures without downtime."