Consistency Spectrum
1. 📈 Eventual Consistency (Weakest)
Key Principle: All nodes will eventually converge to the same state, but not immediately.
✨ Characteristics:
Finite Time Convergence: All nodes will eventually reach the same state after a finite period
No Immediate Consistency: If we keep making changes, nodes will never be in the same state
High Availability: System is highly available
Examples
DNS is highly available system follows eventual consistency
Cassandra DB follows this consistency pattern
2. ⚖️ Causal Consistency
Key Principle: Operations that are causally related must be seen in the same order by all nodes.
🎯 How it Works:
Works by categorising operations into:
3. 🔄 Sequential Consistency
Key Principle: All operations should appear to execute in some sequential order, and this order should be consistent across all nodes.
📊 Requirements:
Single Global Order: All operations must appear in the same sequence everywhere
Program Order Preservation: Operations from each process must appear in program order
4. 🔒 Strict Consistency (Linearisable) 💪
Key Principle: The strongest consistency model - operations appear to take effect instantaneously at some point between their start and finish.
⚡ Characteristics:
Real-time Ordering: Respects the real-time order of operations
Atomic Operations: Each operation appears to take effect instantaneously
Synchronous Replication: Changes must be replicated immediately
High Cost: Requires significant coordination overhead
Example : Password change
Why Do We Need Consensus Algorithms?
📋 Raft Consensus-Based Algorithm
🔄 Quorum-Based Replication
Formula: W + R > N
N = Total number of replicas
W = Minimum number of replicas that must acknowledge a write before it’s considered successful
R = Minimum number of replicas that must respond to a read
