Aws Elasticache

Step 3: Generate cluster configuration

Create ElastiCache (CloudFormation/Terraform) with:

Redis:

• Node type (cache.r7g for memory-intensive, cache.m7g for general)
• Cluster mode: disabled (single shard) or enabled (multiple shards)
• Replication group with 1-5 read replicas per shard
• Multi-AZ with automatic failover
• Subnet group (private/isolated subnets)
• Security group (app tier only, port 6379)
• Parameter group (maxmemory-policy, timeout, etc.)
• At-rest encryption (KMS) and in-transit encryption (TLS)
• Auth token or IAM authentication
• Automatic backups with retention
• Maintenance window

Memcached:

• Node type and count
• AZ placement
• Parameter group (max_item_size, chunk_size)

Step 4: Generate caching strategy code

Implement the appropriate pattern:

Cache-Aside (Lazy Loading):

1. Check cache → if hit, return cached data
2. On miss → query database
3. Write result to cache with TTL
4. Return data

Write-Through:

1. Write to cache AND database simultaneously
2. Always consistent but higher write latency

Write-Behind:

1. Write to cache immediately
2. Async write to database (with queue)
3. Fast writes but risk of data loss

Generate client code with:

• Connection pooling
• Retry logic with exponential backoff
• Serialization/deserialization
• TTL management
• Cache key naming conventions

Step 5: Performance tuning

• Eviction policy (volatile-lru, allkeys-lru, volatile-ttl)
• Memory management and reserved memory
• Connection limits
• Slow log monitoring
• CloudWatch alarms (CPU, memory, connections, cache hit ratio)

Best practices:

• Use Redis for most use cases (richer features, persistence)
• Enable cluster mode for > 100GB or high throughput
• Use reader endpoint for read-heavy workloads
• Set appropriate TTLs (don't cache forever)
• Use cache key prefixes for namespacing
• Enable TLS in-transit encryption
• Monitor cache hit ratio (aim for > 80%)
• Use ElastiCache Serverless for variable workloads