Platform Engineering: Golden Paths

Platform engineering primarily aims to enhance developer productivity and streamline adherence to infrastructure and security policies. It achieves this by automating and simplifying development workflows, allowing developers to focus on innovation.

Enter the concept of golden paths — predefined, opinionated workflows that guide developers through various aspects of their work.

The three Categories of Golden Paths

1. Service Onboarding

These focuses on the effortless onboarding of new services/applications onto the infrastructure. By automating these tasks, we avoid the redundancy of teams having to repeatedly solve the same problems with each new service launch.

A few examples of these include

• Pipelines for generating a new code repository with a template folder structure, sample code, tests, kubernetes deployment file and CI/CD pipeline configuration.
• Provisioning cloud resources and deploying services.
• Exposing an API publicly via Akamai.
• Seamlessly publishing application logs and metrics to a centralized monitoring tool.

2. Day Two Operations

Day two operations are routine tasks and maintenance activities that keep services running smoothly after their launch. If the platform does not deliver a well designed self-serve pipeline to handle these, the platform engineering team can very quickly get overwhelmed by support requests.

Examples include

• Patch management
• Configuration management
• Automated database backups
• Easy rollback
• Automated vulnerability scanning
• Automating these processes ensures that systems remain reliable and secure without requiring constant manual oversight.

3. Reporting

The final golden path focuses on reporting — facilitating informed decision-making among stakeholders. It simplifies the collection and visibility of valuable metrics, promoting the sharing of insights. This approach guarantees that data is not merely collected but is also actively leveraged to inform strategic decisions and drive improvements.

Examples include

• Automated dashboards for each service, such as — performance metrics, CPU/RAM usage & service uptime.
• API latencies
• Infrastructure cost breakdown
• Code quality & test coverage
• DORA metrics

Conclusion

Most self-serve pipelines delivered by a platform engineering team fit into these categories. Based on the maturity of the platform, a development team either interacts with several separate pipelines or a single, well-structured workflow to accomplish various tasks as once.

Understanding the specific category a golden path falls under is crucial for platform teams as they prioritize and design these workflows. This knowledge not only aids in identifying which areas will deliver the most immediate benefits to development teams but also ensures the creation of appropriate abstractions that simplify complex processes without sacrificing functionality.