Notes from https://www.hellointerview.com/learn/system-design/core-concepts/caching

• Comes up often
• When database is getting hammered with reads
• Store frequently accessed data in fast memory to skip the database entirely for most reads
  • Cache hit on Redis takes 1ms
  • Typical database query 20-50ms
  • 20-50x speed up
  • Reduces load on database
    • Handle more write traffic
    • Avoid scaling prematurely
• Cache-aside with Redis
  • Use 90% of the time
  • On a read, check cache first
  • If data is there, return it
• Caching introduces real complexity
  • Hardest part is invalidation
    • E.g user updates profile in database
      • Need to delete or update cached copy
      • Otherwise next read is stale
    • Strategies
      • Invalidate cache entry immediately after writes
      • Use short TTLs and accept some staleness
      • Combination
      • Choice depends on how fresh data needs to be
  • Cache stampedes
    • When a popular cache entry expires, many concurrent requests all miss at the same time and pile onto the database to regenerate it
      • Can spike database load and take down system
    • Prevention
      • Locking
        • Only one request regenerates the entry while the rest wait
      • Early recomputation
        • Regresh entries before they expire
      • Staggering TTLs
        • Entries don’t all expire at once
  • Full cache outage
    • Redis goes down entirely
    • Every request hits database
    • Defenses
      • Small in-process fallback cache
      • Circuit breakers to shed load
      • Graceful degradation until Redis recovers
• Common mistake
  • Caching everything
  • Cache only data that’s read frequently and doesn’t change often
  • Profile system first, cache the hot paths
• CDN caching
  • For static assets like images, videos, and JavaScript files served from edge locations close to users
• In-process caching
  • Works for small values that change rarely
  • Feature flags and config data
• Redis is default for core application data

Notes for https://www.hellointerview.com/learn/system-design/core-concepts/caching