Cache Invalidator Agent

The Cache Invalidator Agent handles intelligent cache invalidation strategies to maintain data consistency across the platform.

📋 Overview

Property	Value
Module	src.agents.cache.cache_invalidator_agent
Class	CacheInvalidatorAgent
Author	Nguyen Viet Hoang
Version	1.0.0

🎯 Purpose

Cache invalidation ensures data consistency by:

• Removing stale data when source data changes
• Propagating updates across distributed cache nodes
• Maintaining consistency between cache and database
• Optimizing memory by removing unused entries

🔧 Invalidation Strategies

Time-Based Invalidation (TTL)

# Set TTL during caching
cache.set("camera:CAM_001", data, ttl=300)  # Expires in 5 minutes

Event-Based Invalidation

# Invalidate on data change event
invalidator.on_entity_update("camera:CAM_001")
invalidator.on_entity_delete("camera:CAM_001")

Pattern-Based Invalidation

# Invalidate all matching keys
invalidator.invalidate_pattern("observation:*")
invalidator.invalidate_pattern("camera:CAM_00*")

Dependency-Based Invalidation

# Invalidate dependent caches
invalidator.invalidate_with_dependencies("camera:CAM_001", [
    "observations:CAM_001:*",
    "stats:CAM_001:*"
])

📊 Architecture

┌─────────────────┐
│ Data Change     │
│ Event           │
└────────┬────────┘
         │
         ▼
┌─────────────────┐    ┌─────────────────┐
│   Invalidator   │───▶│  Cache Manager  │
└────────┬────────┘    └─────────────────┘
         │
    ┌────┴────┐
    ▼         ▼
┌───────┐ ┌───────┐
│ Redis │ │ Local │
│ Cache │ │ Cache │
└───────┘ └───────┘

🚀 Usage

Basic Invalidation

from src.agents.cache.cache_invalidator_agent import CacheInvalidatorAgent

invalidator = CacheInvalidatorAgent()

# Invalidate single key
invalidator.invalidate("camera:CAM_001")

# Invalidate multiple keys
invalidator.invalidate_many([
    "camera:CAM_001",
    "camera:CAM_002"
])

# Invalidate by pattern
invalidator.invalidate_pattern("observation:2025-11-*")

Event-Driven Invalidation

# Subscribe to entity changes
invalidator.subscribe_to_changes([
    "Camera",
    "Observation",
    "Accident"
])

# Handle change event
def on_camera_update(entity_id: str):
    invalidator.invalidate(f"camera:{entity_id}")
    invalidator.invalidate_pattern(f"observations:{entity_id}:*")

⚙️ Configuration

# config/cache_config.yaml
cache_invalidation:
  enabled: true
  
  # Invalidation strategies
  strategies:
    time_based:
      default_ttl: 3600
      camera_ttl: 300
      observation_ttl: 60
    
    event_based:
      enabled: true
      entity_types:
        - Camera
        - Observation
        - Accident
        - Congestion
    
    pattern_based:
      batch_size: 1000
      scan_count: 100

📈 Metrics

Metric	Description
invalidations_total	Total invalidation operations
invalidations_by_pattern	Pattern-based invalidations
invalidation_latency	Average invalidation time
cascade_invalidations	Dependency cascade operations

🛡️ Best Practices

1. Use Specific Keys

# Good: Specific key
invalidator.invalidate("camera:CAM_001:location")

# Avoid: Broad pattern
invalidator.invalidate_pattern("camera:*")

2. Batch Invalidations

# Batch for better performance
invalidator.invalidate_many(key_list)

3. Log Invalidations

# Enable audit logging
invalidator.set_audit_logging(True)

📖 Related Documentation

• Cache Manager Agent - Primary caching operations
• State Updater Agent - State change handling
• Entity Publisher Agent - Entity updates

---

See the complete agents reference for all available agents.