Build_abstractions__not_illusions___PlatformCon_2023_4

Introduction to Abstractions

• Importance of abstractions in computer systems.

• Modern technology is complex; platforms like AWS provide powerful features but entail complexity.

• Need for layers of indirection to simplify interactions with complex systems.

Layers of Abstraction

True vs. Poor Abstraction

• Example of poor abstraction: "piston crankshaft gear wheel assembly builder singleton."

• True abstraction emphasizes functionality (e.g., "automobile" vs. parts).

Composition vs. Abstraction

• Composition combines multiple elements but does not hide details (e.g., Solution Constructs from AWS).

• Effective abstraction shields users from underlying complexity, offering a new vocabulary.

Characteristics of Good Abstractions

• A good abstraction simplifies complex systems (e.g., gas pedal vs. car's internal mechanics).

• Example: Socket streams abstract network complexities.

Leaky Abstractions

• Joel Spolsky's "Law of Leaky Abstractions": all abstractions have limitations.

• Leaks occur from underlying physical principles (e.g., latency, packet loss).

• Not all complexities can be abstracted away; aware of potential limitations.

Distinction Between Abstraction and Illusion

Essentials vs. Non-Essentials

• Abstractions must balance what details to include/exclude.

• Dangerous illusion example: RPC (Remote Procedure Call) misleads users into thinking it operates like local calls.

• RPC introduces issues like latency and failures, obscuring real complexities.

Key Considerations in Abstraction Design

Important Aspects to Consider

• Push vs. Pull: Control over event flow.

• Transaction scopes and consistency in distributed systems.

• Handling timeouts and retries responsibly is crucial.

Less Relevant Details

• Specific cloud API details may be less essential depending on user needs.

• Understanding service interactions is important for cost considerations.

Creating Effective Abstraction Frameworks

• Use coherent language for abstraction (e.g., event-driven systems).

• A cohesive vocabulary structure organizes thought and communication.

• Manage control flows visibly to prevent it from becoming an illusion.

Conclusion

• Abstractions significantly reduce cognitive load for users.

• Effective abstractions have structured languages and must not omit critical details.

• Avoid making oversimplified abstractions that mislead users.

• Insightful approach leads to a more satisfactory user experience.