Consistency Model Hierarchy
1
Strict Serializable
Strongest Guarantee
↓
2
Linearizable
Real-time Atomic
2
Serializable
Transaction Order
↓
3
Sequential
Global Order
3
Snapshot Isolation
Consistent Snapshots
3
Repeatable Read
Same Read Results
↓
4
Causal
Cause-Effect Order
4
PRAM
Per-Process Order
4
Read Committed
Only Committed Data
↓
5
Session Guarantees
Per-Session Consistency
↓
6
Eventual
Eventually Consistent
Detailed Relationship Diagram
graph TD
SS[Strict Serializable
🔒 Banking Systems] --> S[Serializable
📋 ACID Databases] SS --> L[Linearizable
⏰ Configuration Systems] S --> RR[Repeatable Read
📖 SQL Isolation] S --> SI[Snapshot Isolation
📸 MVCC Databases] L --> SEQ[Sequential Consistency
🔄 Distributed Caches] RR --> RC[Read Committed
✅ Web Applications] SI --> RC SEQ --> CAUSAL[Causal Consistency
🔗 Social Media] CAUSAL --> PRAM[PRAM Consistency
🚰 Per-Process Order] CAUSAL --> WFR[Writes Follow Reads
📝 Document Editing] PRAM --> MR[Monotonic Reads
📈 Never Go Backward] PRAM --> MW[Monotonic Writes
📤 Ordered Updates] PRAM --> RYW[Read Your Writes
👁️ See Own Changes] RC --> RU[Read Uncommitted
⚠️ Analytics Systems] MR --> EC[Eventual Consistency
🕐 DNS, CDN] MW --> EC RYW --> EC WFR --> RYW classDef strongest fill:#fed7d7,stroke:#e53e3e,stroke-width:3px,color:#742a2a classDef strong fill:#fbb6ce,stroke:#d53f8c,stroke-width:2px,color:#702459 classDef medium fill:#fbd38d,stroke:#dd6b20,stroke-width:2px,color:#7b341e classDef weak fill:#c3dafe,stroke:#5a67d8,stroke-width:2px,color:#3c366b classDef weakest fill:#b2f5ea,stroke:#38b2ac,stroke-width:2px,color:#234e52 class SS strongest class S,L strong class RR,SI,SEQ medium class CAUSAL,PRAM,RC,WFR weak class MR,MW,RYW,RU,EC weakest
🔒 Banking Systems] --> S[Serializable
📋 ACID Databases] SS --> L[Linearizable
⏰ Configuration Systems] S --> RR[Repeatable Read
📖 SQL Isolation] S --> SI[Snapshot Isolation
📸 MVCC Databases] L --> SEQ[Sequential Consistency
🔄 Distributed Caches] RR --> RC[Read Committed
✅ Web Applications] SI --> RC SEQ --> CAUSAL[Causal Consistency
🔗 Social Media] CAUSAL --> PRAM[PRAM Consistency
🚰 Per-Process Order] CAUSAL --> WFR[Writes Follow Reads
📝 Document Editing] PRAM --> MR[Monotonic Reads
📈 Never Go Backward] PRAM --> MW[Monotonic Writes
📤 Ordered Updates] PRAM --> RYW[Read Your Writes
👁️ See Own Changes] RC --> RU[Read Uncommitted
⚠️ Analytics Systems] MR --> EC[Eventual Consistency
🕐 DNS, CDN] MW --> EC RYW --> EC WFR --> RYW classDef strongest fill:#fed7d7,stroke:#e53e3e,stroke-width:3px,color:#742a2a classDef strong fill:#fbb6ce,stroke:#d53f8c,stroke-width:2px,color:#702459 classDef medium fill:#fbd38d,stroke:#dd6b20,stroke-width:2px,color:#7b341e classDef weak fill:#c3dafe,stroke:#5a67d8,stroke-width:2px,color:#3c366b classDef weakest fill:#b2f5ea,stroke:#38b2ac,stroke-width:2px,color:#234e52 class SS strongest class S,L strong class RR,SI,SEQ medium class CAUSAL,PRAM,RC,WFR weak class MR,MW,RYW,RU,EC weakest
Key Concepts
Consistency vs Availability
Stronger consistency models provide better guarantees but limit system availability during network partitions. This fundamental trade-off shapes distributed system design.
Performance Impact
Higher consistency levels typically require more coordination between nodes, resulting in increased latency and reduced throughput.
Model Relationships
Consistency models form a hierarchy where stronger models imply weaker ones. Understanding these relationships helps in system design decisions.
Real-World Applications
Different parts of your system may need different consistency levels. Financial transactions need strong consistency, while user preferences can use weaker models.
Start Your Deep Dive
Explore comprehensive explanations, interactive demos, and practical examples