Serialized LOB Pattern — Store Object Graphs as Large Objects

In this tutorial, you'll learn how the Serialized LOB pattern stores complex object graphs as serialized large objects in a single database column.

What You'll Learn

how the Serialized LOB pattern stores complex object graphs as serialized large objects in a single database column.

Why It Matters

Mapping complex object graphs to normalized tables is expensive. Serialized LOB is simpler for internal data.

Real-World Use

Hibernate serialized LOB types, PostgreSQL JSONB columns, and MongoDB documents use this approach.

The Serialized LOB Pattern

The Serialized LOB 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

• Registry/Tracking: Serialized LOB maintains a registry of objects or operations.
• Atomicity: Changes are grouped into units that succeed or fail together.
• Isolation: Each unit operates independently.
• Consistency: The pattern ensures data integrity across operations.

Structure

The following diagram shows the structure of this pattern:

classDiagram
    class SerializedLOB {
        -new: List
        -dirty: List
        -removed: List
        +registerNew()
        +registerDirty()
        +registerRemoved()
        +commit()
    }
    class Entity { id data }
    class DataMapper { +insert() +update() +delete() }
    SerializedLOB --> Entity : tracks
    SerializedLOB --> DataMapper : persists

Implementation

from abc import ABC, abstractmethod
from dataclasses import dataclass, field
from typing import List, Dict

@dataclass
class Entity:
    id: int
    data: str = ""

class SerializedLOBRegistry:
    def __init__(self):
        self._new: List[Entity] = []
        self._dirty: List[Entity] = []
        self._removed: List[Entity] = []

    def register_new(self, e: Entity):
        self._new.append(e)

    def register_dirty(self, e: Entity):
        if e not in self._dirty:
            self._dirty.append(e)

    def register_removed(self, e: Entity):
        self._removed.append(e)

    def commit(self):
        print(f"Inserting {len(self._new)} new entities")
        print(f"Updating {len(self._dirty)} dirty entities")
        print(f"Deleting {len(self._removed)} removed entities")
        self._new.clear()
        self._dirty.clear()
        self._removed.clear()

# Usage
reg = SerializedLOBRegistry()
e1 = Entity(1, "Alice")
e2 = Entity(2, "Bob")
reg.register_new(e1)
reg.register_new(e2)
e1.data = "Alice Updated"
reg.register_dirty(e1)
reg.register_removed(e2)
reg.commit()

Expected output:

Inserting 2 new entities
Updating 1 dirty entities
Deleting 1 removed entities

Key Participants

• Serialized LOB: Coordinates tracking and persistence of changes.
• Entity: The domain object being tracked.
• Client: Code that uses the Serialized LOB.
• Data Mapper: Handles actual database operations.

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

• Data Mapper

• Unit Of Work

• Embedded Value

• Design Patterns — the complete patterns catalog.

Pros and Cons

Common Mistakes

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

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

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

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

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

Practice Questions

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

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

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

4. How would you refactor legacy code to introduce Serialized LOB?

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

Challenge

Implement a complete Serialized LOB 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 Serialized LOB pattern. Refactor it, write tests, and measure the improvement in testability, coupling, and cohesion.

Security Tip: When implementing Serialized LOB, 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.