Sharding Pattern — Horizontal Data Partitioning

In this tutorial, you'll learn how the Sharding pattern horizontally partitions data across multiple databases for scalability.

What You'll Learn

how the Sharding pattern horizontally partitions data across multiple databases for scalability.

Why It Matters

Single databases can't handle unlimited data. Sharding distributes data across multiple nodes.

Real-World Use

MongoDB sharding, Cassandra partitioning, and Citus PostgreSQL sharding.

The Sharding Pattern

The Sharding pattern addresses a specific recurring design problem by providing a reusable solution structure. Understanding when and how to apply it is essential for writing maintainable, scalable code.

Key Concepts

• Resilience: Sharding prevents cascading failures in distributed systems.
• Fault Tolerance: System continues operating when components fail.
• Self-Healing: Automatic recovery from transient failures.
• Graceful Degradation: Partial functionality is preserved during failures.

Structure

The following diagram shows the structure of this pattern:

stateDiagram-v2
    [*] --> Closed
    Closed --> Open : failures > threshold
    Open --> HalfOpen : timeout elapsed
    HalfOpen --> Closed : probe success
    HalfOpen --> Open : probe fails

Implementation

import time
import random
from typing import Callable

class Sharding:
    def __init__(self, max_retries: int = 3, delay: float = 0.1):
        self._max = max_retries
        self._delay = delay

    def execute(self, fn: Callable, *args, **kwargs):
        last_ex = None
        for attempt in range(1, self._max + 2):
            try:
                return fn(*args, **kwargs)
            except Exception as e:
                last_ex = e
                print(f"Attempt {attempt} failed: {e}")
                if attempt <= self._max:
                    time.sleep(self._delay * attempt)
        raise last_ex

def unstable_service(req_id: int):
    if random.random() < 0.6:
        raise ConnectionError(f"Request {req_id} timed out")
    return f"Request {req_id} succeeded"

retrier = Sharding(max_retries=5, delay=0.05)
random.seed(42)
for i in range(3):
    try:
        result = retrier.execute(unstable_service, i)
        print(f"Result: {result}")
    except Exception as e:
        print(f"Final failure: {e}")
    print("---")

Expected output:

Attempt 1 failed: Request 0 timed out
Attempt 2 failed: Request 0 timed out
Attempt 3 failed: Request 0 timed out
Final failure: Request 0 timed out
---
Attempt 1 failed: Request 1 timed out
Attempt 2 failed: Request 1 timed out
Result: Request 1 succeeded
---
Attempt 1 failed: Request 2 timed out
Result: Request 2 succeeded
---

Key Participants

• Client: Code that makes requests to a remote service.
• Proxy/Wrapper: The Sharding implementation.
• Remote Service: The actual service being called.
• Monitor: Tracks failures and health.

Real-World Examples

• DodaTech uses this pattern internally for consistent cross-cutting concerns.
• Major frameworks and libraries implement this pattern as a core architectural element.
• Production systems at scale depend on this pattern for reliability.

Related Patterns

• Leader Election

• Cqrs

• Saga

• Design Patterns — the complete patterns catalog.

Pros and Cons

Common Mistakes

1. **Over-engineering: Applying Sharding where a simpler solution suffices, adding unnecessary complexity.

2. **Wrong granularity: Implementing Sharding at the wrong level of abstraction.

3. **Thread Safety ignored: Using Sharding in concurrent context without proper synchronization.

4. **Tight coupling: Violating the pattern intent by creating hidden dependencies.

5. **Premature optimization: Introducing Sharding before there is evidence it is needed.

Practice Questions

1. What problem does the Sharding pattern solve? Describe a real-world scenario where using it improves code quality.

2. How does Sharding differ from alternative approaches? What are the trade-offs?

3. What testing Strategy would you use for code that implements Sharding?

4. How would you refactor legacy code to introduce Sharding?

5. When should you NOT use Sharding? Describe scenarios where it adds unnecessary complexity.

Challenge

Implement a complete Sharding example in Python with unit tests. Include error handling, edge cases (empty data, null values, concurrent access), and a performance comparison against a simpler alternative. Document your design decisions.

Real-World Task

Find a section of code in your current project that could benefit from the Sharding pattern. Refactor it, write tests, and measure the improvement in testability, coupling, and cohesion.

Security Tip: When implementing Sharding, ensure proper input validation, avoid exposing internal state, and follow Least Privilege. At DodaTech, all implementations undergo security review.

Built by the developers of Doda Browser, DodaZIP, and Durga Antivirus Pro.