diff --git a/adr/0022-rate-limiting-cache-strategy.md b/adr/0022-rate-limiting-cache-strategy.md
new file mode 100644
index 0000000..c37d9ab
--- /dev/null
+++ b/adr/0022-rate-limiting-cache-strategy.md
@@ -0,0 +1,536 @@
+# ADR 0022: Rate Limiting and Cache Strategy
+
+## Status
+**Proposed** 🟡
+
+## Context
+
+As the dance-lessons-coach application grows and potentially serves multiple users simultaneously, we need to implement rate limiting to:
+
+1. **Prevent abuse** of API endpoints
+2. **Protect against DDoS attacks**
+3. **Ensure fair usage** across all users
+4. **Maintain system stability** under load
+5. **Provide consistent performance**
+
+Additionally, we need a caching strategy to:
+1. **Reduce database load** for frequently accessed data
+2. **Improve response times** for common requests
+3. **Support horizontal scaling** with shared cache
+4. **Handle cache invalidation** properly
+
+## Decision
+
+We will implement a **multi-phase caching and rate limiting strategy** with the following components:
+
+### Phase 1: In-Memory Cache with TTL Support
+
+**Library Selection**: We will use **`github.com/patrickmn/go-cache`** for in-memory caching because:
+
+✅ **Pros:**
+- Simple, lightweight, and well-maintained
+- Built-in TTL (Time-To-Live) support
+- Thread-safe by default
+- No external dependencies
+- Good performance for single-instance applications
+- Supports automatic expiration
+
+❌ **Cons:**
+- Not shared between multiple instances
+- Memory-bound (not persistent)
+- Limited advanced features
+
+**Implementation Plan:**
+```go
+type CacheService interface {
+    Set(key string, value interface{}, expiration time.Duration) error
+    Get(key string) (interface{}, bool)
+    Delete(key string) error
+    Flush() error
+    GetWithTTL(key string) (interface{}, time.Duration, bool)
+}
+
+type InMemoryCacheService struct {
+    cache *cache.Cache
+    defaultTTL time.Duration
+    cleanupInterval time.Duration
+}
+```
+
+**Use Cases:**
+- JWT token validation results
+- User session data
+- Frequently accessed greet messages
+- API response caching for idempotent endpoints
+
+### Phase 2: Redis-Compatible Shared Cache
+
+**Library Selection**: We will use **`github.com/redis/go-redis/v9`** with a **Redis-compatible open-source alternative**:
+
+**Primary Choice**: **Dragonfly** (https://www.dragonflydb.io/)
+- Redis-compatible
+- Open-source (Apache 2.0 license)
+- Written in C++ with multi-threaded architecture
+- 25x higher throughput than Redis
+- Lower latency
+- Drop-in Redis replacement
+
+**Fallback Choice**: **KeyDB** (https://keydb.dev/)
+- Multi-threaded Redis fork
+- Open-source (GPL license)
+- Better performance than Redis
+- Full Redis API compatibility
+
+**Implementation Plan:**
+```go
+type RedisCacheService struct {
+    client *redis.Client
+    defaultTTL time.Duration
+    prefix string
+}
+
+func NewRedisCacheService(config *config.CacheConfig) (*RedisCacheService, error) {
+    client := redis.NewClient(&redis.Options{
+        Addr:     config.Host + ":" + strconv.Itoa(config.Port),
+        Password: config.Password,
+        DB:       config.Database,
+        PoolSize: config.PoolSize,
+    })
+    
+    // Test connection
+    _, err := client.Ping(context.Background()).Result()
+    if err != nil {
+        return nil, fmt.Errorf("failed to connect to Redis: %w", err)
+    }
+    
+    return &RedisCacheService{
+        client: client,
+        defaultTTL: config.DefaultTTL,
+        prefix: config.Prefix,
+    }, nil
+}
+```
+
+**Configuration:**
+```yaml
+cache:
+  # In-memory cache configuration
+  in_memory:
+    enabled: true
+    default_ttl: 5m
+    cleanup_interval: 10m
+    max_items: 10000
+  
+  # Redis-compatible cache configuration
+  redis:
+    enabled: false
+    host: "localhost"
+    port: 6379
+    password: ""
+    database: 0
+    pool_size: 10
+    default_ttl: 5m
+    prefix: "dlc:"
+    use_dragonfly: true  # Set to false to use KeyDB
+```
+
+### Phase 3: Rate Limiting Implementation
+
+**Library Selection**: We will use **`github.com/ulule/limiter/v3`** because:
+
+✅ **Pros:**
+- Multiple storage backends (in-memory, Redis, etc.)
+- Sliding window algorithm
+- Distributed rate limiting support
+- Configurable rate limits
+- Middleware support for Chi router
+- Good performance
+
+**Implementation Plan:**
+```go
+// Rate limit configuration
+type RateLimitConfig struct {
+    Enabled          bool          `mapstructure:"enabled"`
+    RequestsPerHour  int           `mapstructure:"requests_per_hour"`
+    BurstLimit        int           `mapstructure:"burst_limit"`
+    IPWhitelist       []string      `mapstructure:"ip_whitelist"`
+    EndpointSpecific  map[string]struct {
+        RequestsPerHour int `mapstructure:"requests_per_hour"`
+        BurstLimit      int `mapstructure:"burst_limit"`
+    } `mapstructure:"endpoint_specific"`
+}
+
+// Rate limiter service
+type RateLimiterService struct {
+    limiter *limiter.Limiter
+    store   limiter.Store
+    config  *RateLimitConfig
+}
+
+func NewRateLimiterService(config *RateLimitConfig) (*RateLimiterService, error) {
+    var store limiter.Store
+    
+    // Use Redis if available, otherwise use in-memory
+    if config.UseRedis {
+        // Initialize Redis store
+        store, err = limiter.NewStoreRedisWithOptions(&limiter.StoreOptions{
+            Prefix: config.RedisPrefix,
+            // ... other Redis options
+        })
+    } else {
+        // Use in-memory store
+        store = limiter.NewStoreMemory()
+    }
+    
+    if err != nil {
+        return nil, fmt.Errorf("failed to create rate limiter store: %w", err)
+    }
+    
+    // Create rate limiter
+    rate := limiter.Rate{
+        Period: time.Hour,
+        Limit:  int64(config.RequestsPerHour),
+    }
+    
+    return &RateLimiterService{
+        limiter: limiter.New(store, rate),
+        store:   store,
+        config:  config,
+    }, nil
+}
+```
+
+**Chi Middleware:**
+```go
+func RateLimitMiddleware(limiter *RateLimiterService) func(http.Handler) http.Handler {
+    return func(next http.Handler) http.Handler {
+        return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+            // Skip rate limiting for whitelisted IPs
+            clientIP := r.Header.Get("X-Real-IP")
+            if clientIP == "" {
+                clientIP = r.RemoteAddr
+            }
+            
+            for _, allowedIP := range limiter.config.IPWhitelist {
+                if clientIP == allowedIP {
+                    next.ServeHTTP(w, r)
+                    return
+                }
+            }
+            
+            // Get rate limit context
+            context, err := limiter.limiter.Get(r.Context(), clientIP)
+            if err != nil {
+                log.Error().Err(err).Str("ip", clientIP).Msg("Rate limit error")
+                http.Error(w, "Internal server error", http.StatusInternalServerError)
+                return
+            }
+            
+            // Check if rate limit is exceeded
+            if context.Reached > 0 {
+                w.Header().Set("X-RateLimit-Limit", strconv.Itoa(limiter.config.RequestsPerHour))
+                w.Header().Set("X-RateLimit-Remaining", "0")
+                w.Header().Set("X-RateLimit-Reset", strconv.Itoa(int(context.Reset)))
+                
+                http.Error(w, "Too many requests", http.StatusTooManyRequests)
+                return
+            }
+            
+            // Set rate limit headers
+            w.Header().Set("X-RateLimit-Limit", strconv.Itoa(limiter.config.RequestsPerHour))
+            w.Header().Set("X-RateLimit-Remaining", strconv.Itoa(limiter.config.RequestsPerHour-int(context.Reached)))
+            w.Header().Set("X-RateLimit-Reset", strconv.Itoa(int(context.Reset)))
+            
+            next.ServeHTTP(w, r)
+        })
+    }
+}
+```
+
+### Phase 4: Cache Invalidation Strategy
+
+**Approach**: Hybrid cache invalidation with multiple strategies:
+
+1. **Time-Based Expiration (TTL)**
+   - All cache entries have a TTL
+   - Automatic expiration prevents stale data
+   - Default TTL: 5 minutes for most data
+
+2. **Event-Based Invalidation**
+   - Cache keys are invalidated on specific events
+   - Example: User data cache invalidated on user update
+   - Uses pub/sub pattern for distributed invalidation
+
+3. **Versioned Cache Keys**
+   - Cache keys include data version
+   - When data changes, version increments
+   - Old cache entries naturally expire
+
+4. **Write-Through Caching**
+   - Data written to database and cache simultaneously
+   - Ensures cache is always up-to-date
+   - Used for critical data that must be consistent
+
+**Cache Key Strategy:**
+```go
+func GetCacheKey(prefix, entityType, entityID string) string {
+    return fmt.Sprintf("%s:%s:%s", prefix, entityType, entityID)
+}
+
+// Example: "dlc:user:123"
+// Example: "dlc:jwt:validation:token_hash"
+```
+
+## Implementation Phases
+
+### Phase 1: In-Memory Cache (Current Sprint)
+- ✅ Research and select in-memory cache library
+- ✅ Implement cache interface and in-memory service
+- ✅ Add cache configuration to config package
+- ✅ Implement basic cache operations (set, get, delete)
+- ✅ Add TTL support and automatic cleanup
+- ✅ Cache JWT validation results
+- ✅ Add cache metrics and monitoring
+
+### Phase 2: Redis-Compatible Cache (Next Sprint)
+- ✅ Set up Dragonfly/KeyDB in development environment
+- ✅ Implement Redis cache service
+- ✅ Add configuration for Redis connection
+- ✅ Implement cache fallback strategy (Redis → in-memory)
+- ✅ Add health checks for Redis connection
+- ✅ Implement distributed cache invalidation
+
+### Phase 3: Rate Limiting (Following Sprint)
+- ✅ Research and select rate limiting library
+- ✅ Implement rate limiter service
+- ✅ Add rate limit configuration
+- ✅ Implement Chi middleware for rate limiting
+- ✅ Add rate limit headers to responses
+- ✅ Implement IP whitelisting
+- ✅ Add endpoint-specific rate limits
+
+### Phase 4: Advanced Features (Future)
+- ✅ Cache warming for critical data
+- ✅ Two-level caching (Redis + in-memory)
+- ✅ Cache compression for large objects
+- ✅ Rate limit exemptions for admin users
+- ✅ Dynamic rate limit adjustment
+- ✅ Cache analytics and usage patterns
+
+## Configuration
+
+```yaml
+# Cache configuration
+cache:
+  in_memory:
+    enabled: true
+    default_ttl: "5m"
+    cleanup_interval: "10m"
+    max_items: 10000
+  
+  redis:
+    enabled: false
+    host: "localhost"
+    port: 6379
+    password: ""
+    database: 0
+    pool_size: 10
+    default_ttl: "5m"
+    prefix: "dlc:"
+    use_dragonfly: true
+
+# Rate limiting configuration
+rate_limiting:
+  enabled: true
+  requests_per_hour: 1000
+  burst_limit: 100
+  ip_whitelist:
+    - "127.0.0.1"
+    - "::1"
+  endpoint_specific:
+    "/api/v1/auth/login":
+      requests_per_hour: 100
+      burst_limit: 10
+    "/api/v1/auth/register":
+      requests_per_hour: 50
+      burst_limit: 5
+```
+
+## Monitoring and Metrics
+
+**Cache Metrics:**
+- Cache hit/miss ratio
+- Average cache latency
+- Cache size and memory usage
+- Eviction rate
+- TTL distribution
+
+**Rate Limit Metrics:**
+- Requests allowed vs rejected
+- Rate limit exceeded events
+- Top limited IPs
+- Endpoint-specific rate limit usage
+
+**Prometheus Metrics:**
+```go
+var (
+    cacheHits = prometheus.NewCounterVec(prometheus.CounterOpts{
+        Name: "cache_hits_total",
+        Help: "Number of cache hits",
+    }, []string{"cache_type", "entity_type"})
+    
+    cacheMisses = prometheus.NewCounterVec(prometheus.CounterOpts{
+        Name: "cache_misses_total",
+        Help: "Number of cache misses",
+    }, []string{"cache_type", "entity_type"})
+    
+    rateLimitExceeded = prometheus.NewCounterVec(prometheus.CounterOpts{
+        Name: "rate_limit_exceeded_total",
+        Help: "Number of rate limit exceeded events",
+    }, []string{"endpoint", "ip"})
+)
+```
+
+## Security Considerations
+
+1. **Cache Security:**
+   - Never cache sensitive user data (passwords, tokens)
+   - Use separate cache prefixes for different data types
+   - Implement cache key hashing for sensitive data
+   - Set appropriate TTLs to limit exposure
+
+2. **Rate Limit Security:**
+   - Prevent rate limit bypass attacks
+   - Use X-Real-IP header for proper IP detection
+   - Implement rate limit for authentication endpoints
+   - Log rate limit violations for security monitoring
+
+3. **Redis Security:**
+   - Use authentication if enabled
+   - Implement TLS for Redis connections
+   - Use separate database numbers for different environments
+   - Limit Redis commands to prevent abuse
+
+## Performance Considerations
+
+1. **Cache Performance:**
+   - Benchmark cache operations
+   - Monitor cache latency
+   - Optimize cache key size
+   - Use appropriate data structures
+
+2. **Rate Limit Performance:**
+   - Use efficient rate limiting algorithm
+   - Minimize middleware overhead
+   - Cache rate limit decisions
+   - Batch rate limit checks where possible
+
+3. **Memory Management:**
+   - Set reasonable cache size limits
+   - Monitor memory usage
+   - Implement cache eviction policies
+   - Use memory-efficient data structures
+
+## Migration Strategy
+
+### From No Cache to In-Memory Cache
+1. Implement cache interface and in-memory service
+2. Add cache configuration with sensible defaults
+3. Gradually add caching to critical endpoints
+4. Monitor cache performance and hit ratios
+5. Adjust TTLs based on usage patterns
+
+### From In-Memory to Redis Cache
+1. Set up Dragonfly/KeyDB in development
+2. Implement Redis cache service
+3. Add fallback logic (Redis → in-memory)
+4. Test with both caches enabled
+5. Gradually migrate to Redis-only
+6. Monitor distributed cache performance
+
+### From No Rate Limiting to Rate Limiting
+1. Implement rate limiter with generous limits
+2. Add monitoring for rate limit events
+3. Gradually tighten limits based on usage
+4. Add IP whitelist for critical services
+5. Implement endpoint-specific limits
+6. Monitor and adjust as needed
+
+## Alternatives Considered
+
+### Cache Libraries
+1. **`github.com/bluele/gcache`** - More features but more complex
+2. **`github.com/allegro/bigcache`** - High performance but no TTL
+3. **`github.com/coocood/freecache`** - Very fast but limited API
+
+### Redis Alternatives
+1. **Redis Enterprise** - Commercial, not open-source
+2. **Memcached** - No persistence, simpler protocol
+3. **Couchbase** - More complex, document-oriented
+
+### Rate Limiting Libraries
+1. **`golang.org/x/time/rate`** - Simple but no distributed support
+2. **`github.com/juju/ratelimit`** - Good but limited features
+3. **Custom implementation** - Too much development effort
+
+## Success Metrics
+
+1. **Cache Effectiveness:**
+   - Cache hit ratio > 80%
+   - Average cache latency < 1ms
+   - Memory usage within limits
+
+2. **Rate Limiting Effectiveness:**
+   - < 1% of legitimate requests blocked
+   - Effective protection against abuse
+   - No impact on normal usage patterns
+
+3. **System Stability:**
+   - Reduced database load by 50%
+   - Consistent response times under load
+   - No cache-related outages
+
+## Risks and Mitigations
+
+| Risk | Mitigation |
+|------|------------|
+| Cache stampede | Implement cache warming and fallback logic |
+| Memory exhaustion | Set reasonable cache size limits and monitor usage |
+| Redis failure | Implement fallback to in-memory cache |
+| Rate limit false positives | Start with generous limits and monitor |
+| Performance degradation | Benchmark before and after implementation |
+| Cache inconsistency | Use appropriate invalidation strategies |
+
+## Future Enhancements
+
+1. **Cache Pre-warming** - Load frequently used data at startup
+2. **Two-Level Caching** - Local cache + distributed cache
+3. **Cache Compression** - For large cache objects
+4. **Dynamic Rate Limits** - Adjust based on system load
+5. **User-Specific Rate Limits** - Different limits for different user tiers
+6. **Cache Analytics** - Detailed usage patterns and optimization
+
+## References
+
+- [go-cache documentation](https://github.com/patrickmn/go-cache)
+- [Dragonfly documentation](https://www.dragonflydb.io/docs)
+- [KeyDB documentation](https://keydb.dev/)
+- [limiter/v3 documentation](https://github.com/ulule/limiter)
+- [Chi middleware documentation](https://github.com/go-chi/chi)
+
+## Decision Drivers
+
+1. **Simplicity** - Easy to implement and maintain
+2. **Performance** - Minimal impact on response times
+3. **Scalability** - Support for horizontal scaling
+4. **Reliability** - Graceful degradation on failures
+5. **Open Source** - Preference for open-source solutions
+6. **Community** - Active development and support
+
+## Conclusion
+
+This ADR proposes a comprehensive caching and rate limiting strategy that will significantly improve the performance, scalability, and reliability of the dance-lessons-coach application. The phased approach allows for gradual implementation and testing, minimizing risk while delivering value at each stage.
+
+The combination of in-memory caching for single-instance deployments and Redis-compatible caching for distributed environments provides flexibility for different deployment scenarios. The rate limiting implementation will protect the application from abuse while maintaining a good user experience.
+
+This strategy aligns with our architectural principles of simplicity, performance, and scalability while using well-established open-source technologies with strong community support.
\ No newline at end of file
diff --git a/adr/README.md b/adr/README.md
index 1282e0e..9f0b55f 100644
--- a/adr/README.md
+++ b/adr/README.md
@@ -79,6 +79,8 @@ Chosen option: "[Option 1]" because [justification]
 * [0018-user-management-auth-system.md](0018-user-management-auth-system.md) - User management and authentication system
 * [0019-postgresql-integration.md](0019-postgresql-integration.md) - PostgreSQL database integration
 * [0020-docker-build-strategy.md](0020-docker-build-strategy.md) - Docker Build Strategy: Traditional vs Buildx
+* [0021-jwt-secret-retention-policy.md](0021-jwt-secret-retention-policy.md) - JWT Secret Retention Policy with Configurable TTL and Retention
+* [0022-rate-limiting-cache-strategy.md](0022-rate-limiting-cache-strategy.md) - Rate Limiting and Cache Strategy with Multi-Phase Implementation
 
 ## How to Add a New ADR