Designing A Simple Url Shortening Service A TinyURL Approach - System Design

Requirements

Functional Requirements:

Shorten URL → short link
Redirect short link → original URL
(Optional) expiration, custom alias, analytics

Non-Functional Requirements:

Low latency (10–50ms, read-heavy)
High availability (≥ 99.99%)
Scalable (billions URLs, high QPS)
Durable storage

API Design

POST /shorten
{ long_url, custom_alias?, expires_at? }
→ { short_url }
GET /{short_code}
→ 302 redirect

High-Level Design

Client

CDN

API Gateway (LB, rate limit)

+----------------------+

| |

Shorten Service Redirect Service

| |

ID Generator Cache (Redis)

| |

+---------> Database (sharded)

Analytics: Kafka → OLAP DB

Detailed Component Design

4.1 ID Generation

Use Snowflake ID (distributed, no bottleneck)
Encode → Base62 → short_code
No collision

4.2 Database

urls(
  id BIGINT PK,
  short_code VARCHAR UNIQUE,
  long_url TEXT,
  created_at,
  expires_at
)

Index: short_code
Scaling:
- Read replicas
- Sharding: hash(short_code) % N
- Consistent hashing for rebalancing

4.3 Cache (Redis)

Key: short_code
Value: long_url

Strategy: Write-around (recommended)

Write:
→ DB only
Read:
→ Cache miss → DB → populate cache

TTL (e.g., 24h)
Cache invalidation on update/delete

4.4 Read Flow (Optimized)

CDN (edge cache)
L1: in-memory cache
L2: Redis
DB fallback
Return 302

Handles high QPS efficiently

4.5 Write Flow

Validate URL
Generate ID
Encode Base62
Store DB
Return short URL

Requirements

Functional Requirements:

Shorten URL → short link
Redirect short link → original URL
(Optional) expiration, custom alias, analytics

Non-Functional Requirements:

Low latency (10–50ms, read-heavy)
High availability (≥ 99.99%)
Scalable (billions URLs, high QPS)
Durable storage

API Design

POST /shorten
{ long_url, custom_alias?, expires_at? }
→ { short_url }
GET /{short_code}
→ 302 redirect

High-Level Design

Client

CDN

API Gateway (LB, rate limit)

+----------------------+

| |

Shorten Service Redirect Service

| |

ID Generator Cache (Redis)

| |

+---------> Database (sharded)

Analytics: Kafka → OLAP DB

Detailed Component Design

4.1 ID Generation

Use Snowflake ID (distributed, no bottleneck)
Encode → Base62 → short_code
No collision

4.2 Database

urls(
  id BIGINT PK,
  short_code VARCHAR UNIQUE,
  long_url TEXT,
  created_at,
  expires_at
)

Index: short_code
Scaling:
- Read replicas
- Sharding: hash(short_code) % N
- Consistent hashing for rebalancing

4.3 Cache (Redis)

Key: short_code
Value: long_url

Strategy: Write-around (recommended)

Write:
→ DB only
Read:
→ Cache miss → DB → populate cache

TTL (e.g., 24h)
Cache invalidation on update/delete

4.4 Read Flow (Optimized)

CDN (edge cache)
L1: in-memory cache
L2: Redis
DB fallback
Return 302

Handles high QPS efficiently

4.5 Write Flow

Validate URL
Generate ID
Encode Base62
Store DB
Return short URL