URL Shortener Engine

Appearance

System Architecture & Scaling to 100x ​

Layered Architecture ​

The application adheres strictly to the classic Enterprise Layered Architecture:

mermaid
flowchart LR
    Route[Routes Layer] --> Controller[Controller Layer]
    Controller --> Service[Service Layer Business Logic]
    Service --> DAO[Data Access Object Layer]
    DAO --> DB[(MongoDB Database)]

Scaling to 100x Load: The First Three Bottlenecks ​

  1. Database Read IOPS on Redirections: Under 100,000 QPS, direct MongoDB queries will saturate connection pools. Mitigation: Introduce Redis caching (read-through) with LRU eviction for the hottest 20% of slugs.
  2. Write Lock Contention on Click Counters: Frequent $inc operations on viral links cause document-level lock contention. Mitigation: Buffer click events in Redis or Apache Kafka and write back to MongoDB in batches every 5 seconds.
  3. Network Bandwidth on Gateway: Serving millions of redirects can saturate gateway network cards. Mitigation: Deploy multi-region Cloudflare Workers / Fastly VCL at the edge to perform redirects directly from edge key-value stores.