Transfer Matrix¶

These patterns are so fundamental they appear in nearly every domain. Recognizing them helps you transfer knowledge across fields.

Indirection / Abstraction¶

Core idea: Adding a layer to gain flexibility

Examples:

• Memory: Virtual memory
• Networking: Protocol layers
• Architecture: Dependency inversion
• Data: Views, indexes

Caching / Memoization¶

Core idea: Trading space for time

Examples:

• Memory: Cache hierarchy
• Distributed: CDNs, replicas
• Performance: Result caching
• Data: Materialized views

Partitioning / Sharding¶

Core idea: Divide to conquer scale

Examples:

• Data: Database sharding
• Distributed: Service decomposition
• Memory: NUMA awareness
• Testing: Test isolation

Replication¶

Core idea: Copies for availability or performance

Examples:

• Data: Database replicas
• Distributed: Service redundancy
• Memory: Cache lines
• Files: Backup systems

Batching¶

Core idea: Amortize overhead across operations

Examples:

• I/O: Buffered writes
• Networking: Request batching
• Performance: Bulk operations
• Data: Batch processing

Queueing¶

Core idea: Decouple producers from consumers

Examples:

• Concurrency: Work queues
• Distributed: Message queues
• I/O: Request queues
• Performance: Connection pools

Why These Patterns Matter¶

These patterns aren't just similar-looking code - they represent deep structural solutions to fundamental problems:

• Indirection solves the problem of coupling
• Caching solves the problem of repeated work
• Partitioning solves the problem of scale
• Replication solves the problem of availability
• Batching solves the problem of overhead
• Queueing solves the problem of coupling in time

When you see a new system or architecture, ask: which of these patterns is it using? This question often reveals the core design decisions and trade-offs.