Welcome to the Caching & Delivery module.
At Staff Engineering levels, caching isn’t just about “putting Redis in front of the DB.” It’s about data consistency models, avoiding thundering herds, and designing for failure.
Module Structure
1. Foundations: Patterns & Topologies
- Strategies: Cache-Aside, Write-Through, Write-Back.
- Topologies: When to use Sidecars vs. Centralized clusters.
- Sharding: Handling Hot Keys and Partitions.
2. Advanced: Consistency & Coordination
- Race Conditions: Why
delete(key)is inherently race-prone. - Leases: How Facebook solves the “Thundering Herd” and Stale Sets.
- CDC: Using the Transaction Log for reliable invalidation.
3. Global Delivery: CDNs & Edge
- Anycast addressing and Edge compute patterns.
- TCP optimization and Dynamic Site Acceleration.
Related Primitives
- Bloom Filters: Probabilistic headers for cache avoidance.
- Consistent Hashing: The algorithm behind sharded caching.
Module Chapters
Caching 101: The Desk Drawer
Why is opening a desk drawer faster than driving to the library? Learn how caching saves time and protects your database from being overwhelmed.
Start Learning →Cache Consistency & Coordination
The hardest problem in computer science. Handling Invalidation Races, Leases, and CDC-based consistency.
Start Learning →Caching Patterns & Topologies
Cache-Aside vs Write-Through, plus Distributed Topologies (Sharded, Sidecar, Replicated).
Start Learning →Bloom Filters: The Probability Guard
How to check if something exists without actually storing it. The magic of hash functions and bit arrays.
Start Learning →Distributed Caching Topologies
Scaling cache beyond a single node. Proxy vs Client-side routing, Hot Key mitigation, and Rehashing storms.
Start Learning →Global Delivery: CDNs & Edge
Global content delivery at the speed of light. Anycast BGP, Split TCP, and the rise of Edge Compute.
Start Learning →