Distributed Systems & Consensus (Raft / Paxos)

आप क्या सीखेंगे

• Explain partial failure, failure and timing models, and why a timeout can only suspect a crash.
• Order events with Lamport and vector clocks instead of unreliable wall clocks.
• Reason about quorums, consistency models, CAP/PACELC, FLP and why 2PC blocks.
• Trace Basic Paxos and Multi-Paxos and argue why quorum intersection keeps them safe.
• Walk through Raft elections, log replication, commit rules, membership changes, snapshots and linearizable reads.
• Use etcd and ZooKeeper for coordination and protect shared resources with leases and fencing tokens.
• Diagnose production pitfalls and choose testing techniques such as deterministic simulation, Jepsen and TLA+.

पाठ्यक्रम

Mental Model: Nodes, Messages and Failure

1. Why Distribute, and Why It Is Hard
2. Failure Models: Crash, Omission and Byzantine
3. Timing Models and Failure Detection

Time, Clocks and Ordering Events

1. Physical Clocks and Why They Lie
2. Lamport Clocks and Happens-Before
3. Vector Clocks and Detecting Concurrency

Replication and Consistency Models

1. Replication Strategies and Quorums
2. Consistency Models: Linearizable to Eventual
3. CAP and PACELC Without the Myths

The Consensus Problem

1. The Consensus Problem and State Machine Replication
2. The FLP Impossibility Result
3. Two-Phase Commit vs Consensus

Paxos

1. Basic Paxos: Prepare, Promise, Accept
2. Why Paxos Is Safe, and Dueling Proposers
3. Multi-Paxos and the Paxos Family

Raft: Terms and Leader Election

1. Raft Overview: Roles, Terms and RPCs
2. Leader Election and Randomised Timeouts

Raft: Log Replication and Safety

1. Log Replication with AppendEntries
2. Committing Entries and the Current-Term Rule
3. Raft Safety: Election Restriction and Leader Completeness

Raft in Practice: Membership, Snapshots and Reads

1. Cluster Membership Changes
2. Snapshots and Log Compaction
3. Linearizable Reads: Log Reads, ReadIndex and Leases

Consensus in Real Systems: etcd, ZooKeeper and Locks

1. etcd: A Raft-Backed Key-Value Store
2. ZooKeeper and the Zab Protocol
3. Distributed Locks, Leases and Fencing Tokens

Performance, Pitfalls and Verification

1. Performance: Latency, Batching and Multi-Raft
2. Production Pitfalls: Split Brain, Pre-Vote and Disruptive Nodes
3. Testing and Verifying Consensus Systems
4. Byzantine Fault Tolerance and Blockchains

Revision and Interview Prep

1. Cheat Sheet: Everything on One Page
2. Common Interview and Exam Questions
3. Mixed Self-Check

संबंधित कोर्स

• C (38 पाठ)
• Cost & Capacity Estimation (30 पाठ)
• EDA & CQRS (25 पाठ)
• HLD & LLD (25 पाठ)