July 2022 Summaries

A Platform Orchestrator is a centerpiece of a dynamic Internal Developer Platform, enabling dynamic configuration management and developer self-service while driving standardization by design. It's an alternate approach to static config management, where a "baking machine" reads a Declarative Application Model and generates a representation of the application with its dependent resources, deploying it dynamically. This approach allows developers to focus on workload specifications without dealing with setup details, and enables standardization by design through workload profiles and resource definitions. The orchestrator aggregates data and logs in one place, applies fine-grained Role Based Access Control, and models applications and dependencies through a CLI or API. It's a powerful feature that acts as an API layer, hooking into almost all parts of the delivery chain, allowing for harmonious orchestration of deployments and infrastructure provisioning, and enabling organizations to move away from static setups.

The author argues that static configuration management, where an entire application and its dependent infrastructure resources are described in a declarative model consistent across any environment, is flawed and leads to errors, hard maintenance, and high standardization costs. The solution proposed is dynamic configuration management, which separates environment-agnostic from environment-specific elements of app configuration, resulting in easier changes, reduced cognitive load for developers, and increased standardization by design. This approach enables the creation of a Declarative Application Model, which dynamically generates the application's representation with every deployment, reducing maintenance overhead, change failure rates, and enabling advanced functionality like self-service for developers and end-to-end audit logs. The author believes that dynamic configuration management will become the predominant approach in the next 10 years due to its benefits on maintainability, standardization, ease of use, and drop-in change failure rate.