Projetando um encurtador de URLs

Welcome! In this lesson, we'll design a URL shortening service, similar to Bitly or TinyURL. These services take a long, complex URL and convert it into a much shorter, more manageable one.

URL shorteners are incredibly useful for sharing links on social media, in emails, or anywhere space is limited. They also often provide analytics, tracking how many times a shortened link is clicked.

A URL shortener primarily performs two key functions:

• Shorten: Takes a long URL as input and generates a unique, short code. This code is then used to construct the short URL.
• Redirect: When a user accesses a short URL, the service looks up the corresponding long URL and redirects the user's browser to it.

These two operations form the backbone of the entire system.

The heart of a URL shortener is its ability to generate unique, short, and often human-readable codes. Common approaches include:

• Sequential IDs + Base62 Encoding: Use an auto-incrementing database ID and convert it to a Base62 string. Base62 uses 0-9, a-z, A-Z (62 characters), allowing for shorter codes than Base10.
• Hash Functions: Apply a hash function (like MD5 or SHA256) to the long URL. Take a portion of the hash to form the short code. This requires collision handling.
• Random String Generation: Generate a random string of a fixed length. This also requires checking for uniqueness to avoid collisions.

Let's look at a simple Python example of Base62 encoding, which is a popular method for generating short codes from sequential IDs. This helps ensure uniqueness while keeping codes compact.

To store our URL mappings, we'll need a database. A simple schema could look like this (using a relational database like PostgreSQL):

• id: Primary key (auto-incrementing integer)
• short_code: VARCHAR(10) - The unique short string
• long_url: TEXT - The original, long URL
• created_at: TIMESTAMP - When the short URL was created
• user_id: INT (optional) - If users can create accounts
• click_count: INT (optional) - For basic analytics

A NoSQL database could also work, offering flexibility for schema evolution.

When a user clicks on a short URL (e.g., https://tiny.url/abcde), the redirection service takes over. It performs these steps:

1. Extracts the short_code (e.g., abcde) from the URL.
2. Queries the database to find the corresponding long_url.
3. Sends an HTTP 301 (Moved Permanently) or 302 (Found) redirect response to the user's browser, pointing to the long_url.

301 vs 302: 301 is for permanent redirects and is cached by browsers, 302 is temporary. For shorteners, 301 is often preferred for performance after the initial creation.

Ensuring each generated short code is unique is critical. If we use hash functions or random strings, collisions (two different long URLs getting the same short code) are possible, though rare with longer codes.

Strategies to handle collisions:

• Database Check: Always attempt to insert the new mapping and catch a unique constraint violation. If a collision occurs, regenerate the code and retry.
• Pre-check: Before inserting, query the database to see if the code already exists. This can lead to race conditions under high concurrency, so database-level unique constraints are preferred.
• Distributed ID Generation: For sequential IDs, use a distributed ID generator (e.g., Snowflake ID) to ensure globally unique IDs that can then be Base62 encoded.

A popular URL shortener needs to handle millions of requests. Key scalability points:

• Database: The database will be a hotspot. Consider sharding the database by short_code or using a distributed key-value store. Read replicas are essential for the redirection service.
• Caching: Cache frequently accessed short URL to long URL mappings (e.g., using Redis or Memcached) to reduce database load, especially for the redirection path.
• Asynchronous Processing: For click analytics, instead of incrementing a counter synchronously, send click events to a message queue for asynchronous processing.
• Load Balancers: Distribute incoming traffic across multiple instances of your shortening and redirection services.

Beyond just shortening, many services offer basic analytics. To track clicks:

• When a short URL is accessed, increment a click_count in the database for that specific mapping.
• For high traffic, this counter update can become a bottleneck. A more scalable approach is to send a message to a queue (e.g., Kafka, RabbitMQ) and have a separate worker process asynchronously update the counts or store detailed click logs.
• Detailed analytics might involve storing referrer, user agent, IP address, etc., in a separate analytics database (e.g., a data warehouse).

We've walked through the core components of designing a URL shortener!

• We covered the two main functions: shortening and redirection.
• Explored methods for generating unique short codes, like Base62 encoding.
• Discussed database schema for storing mappings and the mechanics of the redirection service.
• Addressed crucial aspects like collision handling, scalability with caching and sharding, and basic click analytics.

This case study illustrates how various system design principles come together to build a functional and scalable service.