⚙️ Backend Developer Senior

CAP: Trong distributed system, chỉ có thể đạt 2 trong 3:

• Consistency: Mọi read nhận được data mới nhất
• Availability: Mọi request nhận được response
• Partition tolerance: System hoạt động khi network partition

CP (Consistency): Banking, inventory - data phải chính xác

AP (Availability): Social media, analytics - chấp nhận eventual consistency

Requirements: Shorten URL, redirect, analytics

Components:

• API: POST /shorten, GET /:shortCode (301 redirect)
• Short code: Base62 encode (a-z, A-Z, 0-9)
• Storage: Key-value store (Redis), Database backup
• Scale: Counter service (Zookeeper), pre-generate codes

Collision handling: Check existence, retry với code mới

1. Horizontal (Row-based): Chia rows theo shard key

• Range: user_id 1-1M → shard1, 1M-2M → shard2
• Hash: hash(user_id) % num_shards
• Directory: Lookup table chỉ đến shard

2. Vertical: Chia columns (separate tables)

Challenges: Cross-shard queries, rebalancing, transactions.

Algorithms:

• Token Bucket: Tokens sinh theo thời gian, request lấy token
• Sliding Window: Count requests trong time window
• Leaky Bucket: Queue requests, process với fixed rate

# Redis Sliding Window
def is_allowed(user_id, limit=100, window=60):
    key = f"rate:{user_id}"
    now = time.time()
    
    pipe = redis.pipeline()
    pipe.zremrangebyscore(key, 0, now - window)
    pipe.zadd(key, {str(now): now})
    pipe.zcard(key)
    pipe.expire(key, window)
    _, _, count, _ = pipe.execute()
    
    return count <= limit

Request-Response: Synchronous, simple, nhưng tight coupling.

Event-Driven:

• Loose coupling, scalable, resilient
• Eventual consistency
• Patterns: Event Sourcing, CQRS, Saga

Khi nào dùng Event-Driven?

• Multiple services cần react to same event
• Async processing acceptable
• Audit trail quan trọng (Event Sourcing)

• Load Balancer: Distribute traffic (L4/L7)
• Connection Pooling: Reuse DB connections
• Async I/O: Event loop (Node.js, Go goroutines)
• Horizontal Scaling: Add more servers
• Caching: CDN, Redis, in-memory
• Database: Read replicas, sharding