Cache Optimization Techniques and Design Strategies

1) Cache Benefits and Cost Analysis

Cache accelerates read/write speed and reduces backend load, but introduces data inconsistency, higher code maintenance, and operational costs (e.g., Redis Cluster).

Typical use cases include heavy computation and request acceleration.

2) Cache Update Strategies

Three main strategies:

LRU/LFU/FIFO eviction : used when cache exceeds its memory limit.

Expiration : set TTL so stale data is removed automatically; suitable for data tolerant of short delays.

Active update : notify cache immediately after source data changes, often via messaging.

Comparison of the three strategies is illustrated in the accompanying diagram.

3) Cache Granularity Control

Choosing appropriate cache granularity balances data reuse, memory consumption, and code maintainability. Common stack: Redis for cache, MySQL for storage.

4) Penetration Optimization

Cache penetration occurs when queries for non‑existent keys miss both cache and storage. Solutions:

Cache null objects : store empty placeholders with short TTL to prevent repeated storage hits.

Bloom filter : pre‑filter keys to avoid unnecessary storage lookups.

Diagrams compare null‑object and Bloom‑filter approaches.

5) Bottom‑Hole Optimization

Adding more cache nodes can degrade performance due to increased network calls and connection overhead. Optimizations focus on reducing network round‑trips, e.g., using command pipelining, smart client slot mapping, and parallel I/O.

6) Avalanche Optimization

Cache avalanche happens when the cache becomes unavailable, flooding the storage layer. Prevention includes high‑availability cache design (Redis Sentinel/Cluster), backend rate‑limiting, and failure drills.

7) Hot‑Key Rebuild Optimization

Hot keys can cause massive load during cache miss. Solutions:

Mutex lock : only one thread rebuilds the cache; others wait (implemented with Redis SETNX ).

Never‑expire : logical expiration with background rebuilds, avoiding simultaneous heavy rebuilds.

Code snippets illustrate the mutex lock approach using SETNX and retry logic.