Distributed System¶

The Internet is the biggest distributed system in the world.

Concerns for Remote Services¶

Where are the remote services located? (Service Discovery)
How many instances of the service are there, and how should traffic be distributed among them? (Load Balancing)
What actions should be taken if there are network timeouts, partition failures, or service errors? (Fault Tolerance: e.g., Circuit Breaker, Isolation, Fallback)
How should remote methods, including arguments and return values, be represented? (Communication Protocols: e.g., Serialization and Protocol)
How should permissions for accessing the services be managed? (Authentication & Authorization)
How can communication be secured? (Network Security)
How can consistent results be ensured across multiple service instances? (Distributed Data Consistency)

Eight Classic Fallacies¶

The network is reliable;
Latency is zero;
Bandwidth is infinite;
The network is secure;
Topology doesn't change;
There is one administrator;
Transport cost is zero;
The network is homogeneous.

CAP-Theorem¶

In a distributed system, it is impossible to guarantee the three properties of Consistency, Availability, and Partition Tolerance at exactly the same time.

You can choose two of them for a specific system or scenario, for example:
AP – Domain Name System (DNS).
CA – Relational Database Management System (RDBMS).
CP – Banking Application.

BASE - BA:Basic Available; S:Soft State; E:Eventual Consistency

Distributed Transaction¶

TODO