duyet

Type	Location	Device	Pages	Last Activity
🧠 llm	Unknown	unknown - unknown desktop	/live	5:11:22 PM

This real-time analytics system tracks website visitors using modern web technologies and provides privacy-compliant insights without storing personal information.

Technical Architecture

🔧 Storage & Performance

Deno KV: Edge-distributed key-value database
TTL: 5-minute session expiration for accurate counts
Atomic Operations: Consistent counters with race condition protection
Background Cleanup: Automatic expired session removal

🔄 Real-time Updates

Server-Sent Events: Live data streaming to browsers
10-second intervals: Balanced between accuracy and performance
Auto-reconnection: Resilient connection handling
Middleware tracking: Seamless session management

User Classification

👤 Humans

Real visitors using browsers with typical user-agent patterns and JavaScript capabilities.

🤖 Bots

Search crawlers, monitoring tools, API clients, and automated browsers.

🧠 LLMs

AI language models like GPTBot, ClaudeBot, PerplexityBot, and other LLM crawlers.

Privacy & Data

✅ No personal data: Only aggregate anonymous metrics
✅ Location headers only: Country/region from CDN headers (no IP storage)
✅ GDPR compliant: No tracking cookies, session-based identification
✅ Automatic cleanup: All data expires automatically via TTL

Technical Details

Session Management: Each visitor gets a temporary session ID (5-minute TTL) to track page visits and activity without persistent identification.

Data Structure: Sessions stored as ["live_sessions", sessionId]with atomic counters at ["live_count", "total"] for consistency.

Trend Analysis: Activity aggregated in 5-minute buckets at ["live_trend", timestamp] for historical insights.

Scalability: Distributed KV storage with edge caching enables global deployment while maintaining real-time performance.

Live Visitors

Locations

Activity Trend