ElastiCache · Redis and Memcached

Outcome

• Redis CachingIAM, Network, and the Redis Connection Model
  • Explain why ElastiCache has no public endpoint and what the connection path must be
  • Create a security group with source-SG inbound rule (not CIDR) for Redis port
  • Explain the difference between ElastiCache control plane IAM and data plane AUTH tokens
  • Debug a Redis connection timeout systematically (SG → subnet → NACL → host)
• Redis CachingCluster Setup and Basic Data Structures
  • Use String for cached responses, counters, and idempotency keys with TTL
  • Use Hash for session objects with independently updatable fields
  • Use Sorted Set for leaderboards and sliding-window time series
  • Explain WRONGTYPE error and how to diagnose it with TYPE command
  • Explain why EXPIRE returning 0 is a silent failure
  • Use MGET to batch multiple key reads in one round-trip
• Redis CachingTTL, Eviction, and the Cache-Aside Pattern
  • Implement cache-aside (read → miss → DB → SET with TTL → return) in a Lambda inside a VPC
  • Explain volatile-lru vs allkeys-lru and when to use each
  • Explain why noeviction is wrong for a cache and what happens when memory fills
  • Explain why every key must have a TTL and what happens without one
  • Namespace keys by service to prevent collision on shared Redis
• Redis CachingRedis Pub/Sub and Streams
  • Use SUBSCRIBE/PUBLISH for cache invalidation signals
  • Explain why Pub/Sub messages are lost if no subscriber is listening
  • Use XADD/XREADGROUP/XACK for reliable audit log processing
  • Explain XPENDING and how XCLAIM handles consumer failure
  • Explain when to use Pub/Sub vs Streams vs SQS
• Redis CachingReplication and High Availability
  • Create a replication group with automatic failover and Multi-AZ
  • Read from replica endpoint, write to primary endpoint
  • Trigger a failover and observe the ~30-second window
  • Explain why writes to a replica fail with READONLY
  • Explain what a client without retry logic experiences during failover
• Redis CachingRate Limiting, Distributed Locks, and Monitoring
  • Implement atomic rate limiting with SET NX EX + INCR
  • Implement sliding window rate limiting with Sorted Set
  • Acquire and release a distributed lock with random token and Lua atomic release
  • Explain why lock without TTL causes permanent deadlock
  • Explain why lock without random token allows double-release
  • Create CloudWatch alarms on CacheHitRate and Evictions
  • Interpret INFO memory and INFO replication output