1. Estimating Scale of the System

1.1. Traffic Estimations

• Assume the system shortens about 100 million URLs per month.
• Assume each short URL is accessed 10 times on average.
• This results in 1 billion redirection requests per month.

1.2. URL Storage Estimations

• If we store URLs for 10 years, the system will store:
  100 million × 12 months × 10 years = 12 billion URLs

1.3. Read vs. Write Ratio

• URL shortening is a write operation.
• URL redirection is a read operation.
• Reads are much higher than writes (typically 100:1 ratio).

---

2. Storage Constraints

2.1. URL Storage Size

Each stored URL requires:

 

• Short URL ID: ~7 bytes (Base62 encoding)
• Original URL: ~500 bytes (average)
• Metadata (timestamp, access count, expiration, etc.): ~100 bytes
• Total Storage per URL = ~600 bytes

 

For 12 billion URLs, the storage required is: 12 billion × 600 bytes ≈ 7.2 TB

 

If we add indexing overhead, storage may reach 10 TB over 10 years.

2.2. Caching Requirements

• Frequently accessed URLs should be cached (Redis, Memcached).
• If 20% of URLs account for 80% of traffic, caching 2.4 billion URLs would require:
• Each cached URL entry ~100 bytes
• Total Cache Size ≈ 240 GB (using Redis or Memcached).

---

3. Bandwidth Constraints

3.1. Shortening Requests (Writes)

• 100 million URL shortening requests per month
• ~40 URLs per second

3.2. Redirection Requests (Reads)

• 1 billion redirections per month
• ~4,000 requests per second (RPS) at peak load

3.3. Data Transfer Estimation

Assuming each redirection request is ~0.5 KB (URL lookup, metadata, logs):

 

• Daily Data Transfer:
  4,000 RPS × 0.5 KB × 86,400 sec ≈ 172 GB/day
• Monthly Data Transfer:
  ~5 TB per month

---

4. Performance & Latency Constraints

4.1. URL Shortening Latency

• Shortening should take <100 ms, including:
• Encoding the URL (Base62)
• Storing it in a database
• Returning the short URL to the user

4.2. URL Redirection Latency

• Redirection should be <10 ms for cached URLs.
• <100 ms for database lookups.
• CDNs and edge caching can reduce latency further.

---

5. Constraints & Challenges

5.1. Database Constraints

• A relational database (MySQL, PostgreSQL) may not scale well beyond a few billion URLs.
• A NoSQL solution (Cassandra, DynamoDB, MongoDB) may be better for handling high traffic.

5.2. Scalability Constraints

• The system must handle sudden traffic spikes.
• Horizontal scaling with multiple servers is needed.

5.3. Security Constraints

• Prevent brute force attacks on short URLs.
• Detect and block malicious URLs (phishing, malware).

5.4. Fault Tolerance & Availability

• Replication across data centers ensures uptime.
• Load balancers distribute traffic evenly.

---

6. Summary of Capacity Estimations