Database Caching (Query Cache, Redis)

Frequently accessed data is stored in a fast-access layer (memory) to avoid hitting the database repeatedly.

Caching Layers

Client
  ↓
Application Cache (in-memory, e.g. Redis)
  ↓
Query Cache (inside the DB engine)
  ↓
Buffer Pool (DB caches disk pages in RAM)
  ↓
Disk (actual data)

1. Query Cache (DB-Level)

Built inside the database engine. Stores the result of a SELECT query and reuses it if the same query is fired again.

• Cache hit → return result directly
• Cache miss → execute query → store result → return
• If underlying table changes → cached result is invalidated immediately
• In write-heavy systems, cache is constantly invalidated → useless overhead

Note: MySQL removed query cache in version 8.0. Works well only for read-heavy, rarely updated data.

2. Redis (Application-Level Cache)

In-memory key-value store used as an external caching layer between your application and the database.

App → Check Redis → HIT  → Return data
              → MISS → Query DB → Store in Redis → Return data

SET user:101 <data> EX 3600   -- cache with TTL
GET user:101                   -- cache hit → instant response

Caching Strategies

Strategy	How it Works	Best For
Cache Aside	App checks cache first, loads from DB on miss	Most common, flexible
Write Through	Write to cache and DB simultaneously	Read-heavy, consistency needed
Write Behind	Write to cache first, DB updated later async	Write-heavy systems
Read Through	Cache itself fetches from DB on miss	Simpler app logic

Cache Problems

Problem	Description
Cache Miss Storm	Many requests hit DB simultaneously on a cold cache
Cache Stampede	Cache expires → thousands of requests flood DB at once
Cache Penetration	Requests for non-existent data bypass cache, hammer DB
Stale Data	Cache serves outdated data if not invalidated properly

Redis vs Query Cache

	Query Cache	Redis
Location	Inside DB engine	External server
Flexibility	Limited	Highly flexible
Data Types	Query results only	Strings, lists, sets, hashes, etc.
Invalidation	Automatic (on table change)	Manual / TTL
Scalability	Poor	Excellent