Cache Manager Agent

The Cache Manager Agent provides distributed caching with Redis backend for improved performance across the UIP platform.

📋 Overview

Property	Value
Module	src.agents.cache.cache_manager_agent
Class	CacheManagerAgent
Author	Nguyen Viet Hoang
Version	1.0.0

🎯 Purpose

The Cache Manager Agent manages distributed caching to:

• Reduce database load by caching frequently accessed data
• Improve response times for API requests
• Enable horizontal scaling with shared cache state
• Implement cache invalidation strategies

🔧 Cache Strategies

Write-Through

Updates cache immediately when data is written to the database.

# Data written to DB and cache simultaneously
cache_manager.set("camera:CAM_001", camera_data, ttl=3600)

Write-Behind (Async)

Asynchronous cache updates for better write performance.

# Queue cache update for async processing
cache_manager.set_async("observations:latest", observations)

Cache-Aside (Lazy Loading)

Load data into cache only on cache miss.

# Check cache first, load from DB on miss
data = cache_manager.get_or_load("pattern:P001", loader_func)

📊 Architecture

┌─────────────────┐    ┌─────────────────┐
│   API Request   │───▶│  Cache Manager  │
└─────────────────┘    └────────┬────────┘
                                │
                    ┌───────────┼───────────┐
                    ▼           ▼           ▼
             ┌──────────┐ ┌──────────┐ ┌──────────┐
             │  Redis   │ │  Local   │ │ Database │
             │  Cache   │ │  Cache   │ │ Fallback │
             └──────────┘ └──────────┘ └──────────┘

🚀 Usage

Basic Usage

from src.agents.cache.cache_manager_agent import CacheManagerAgent

# Initialize with configuration
config = {
    "enabled": True,
    "redis_host": "localhost",
    "redis_port": 6379,
    "default_ttl": 3600
}
cache_manager = CacheManagerAgent(config)

# Set value
cache_manager.set("key", {"data": "value"}, ttl=300)

# Get value
value = cache_manager.get("key")

# Delete value
cache_manager.delete("key")

Cache Key Patterns

# Entity caching
cache_manager.set("camera:CAM_001", camera_entity)
cache_manager.set("observation:OBS_001", observation)

# Query result caching
cache_manager.set("query:accidents:severe", results, ttl=60)

# Aggregation caching
cache_manager.set("stats:hourly:2025-11-29T10", hourly_stats)

⚙️ Configuration

Environment Variables

REDIS_HOST=localhost
REDIS_PORT=6379
REDIS_PASSWORD=your_password
REDIS_DB=1

YAML Configuration

# config/cache_config.yaml
cache:
  enabled: true
  redis_host: localhost
  redis_port: 6379
  redis_db: 1
  default_ttl: 3600
  
  # TTL by category
  ttl_settings:
    cameras: 300
    observations: 60
    patterns: 600
    statistics: 3600

📈 Performance Metrics

The Cache Manager tracks:

• Hit Rate: Percentage of successful cache hits
• Miss Rate: Percentage of cache misses
• Latency: Average cache operation latency
• Memory Usage: Redis memory consumption

🔄 Cache Invalidation

Manual Invalidation

# Invalidate single key
cache_manager.delete("camera:CAM_001")

# Invalidate by pattern
cache_manager.delete_pattern("camera:*")

# Clear all cache
cache_manager.flush()

Automatic Invalidation

TTL-based expiration ensures stale data is automatically removed.

🛡️ Fallback Behavior

When Redis is unavailable, the agent falls back to:

1. In-memory cache for local operations
2. Direct database queries for critical data
3. Graceful degradation without errors

📖 Related Documentation

• Cache Invalidator Agent - Cache invalidation strategies
• Performance Monitor - Cache performance tracking
• Redis Configuration - Redis setup

---

See the complete agents reference for all available agents.