Skip to content

Data Models for Metrics

The data models in models.py are designed with a strong focus on metrics, allowing for efficient storage and manipulation of various deployment metrics. These models provide a structured way to represent and work with metric data throughout the Orbis system.

Metric-Focused Design

  1. Metric Class: This is the core model for representing individual metrics. It includes:

    • Metric identifier and name
    • Associated queries
    • Aggregators and table headers
    • Worker queue information
    • Executor type

  2. MetricCalculation Class: Represents calculated values for a specific metric, including:

    • Mean, median, max, min, and 90th percentile values
    • Optional worker queue statistics

  3. WorkerQueueStats Class: Stores detailed statistics for individual worker queues, allowing for granular analysis of queue performance.

  4. Figure Class: Represents a generated visualization of a metric, linking the metric data with its visual representation.

  5. NamespaceReport and OverallReport Classes: These classes aggregate metric calculations for namespaces and the entire organization, respectively.

Flexibility for New Metrics

Adding new metrics to the system is relatively straightforward:

  1. YAML Configuration: Add the new metric queries to the YAML configuration file.

  2. Model Updates: In most cases, no changes to the existing models are required. The Metric class is designed to accommodate various types of metrics.

  3. Processing Logic: Update the processing logic in generator.py and csv_generator.py to handle the new metric if it requires special treatment.

Extensibility

The current model structure allows for easy extension:

  1. New Attributes: If a new metric requires additional attributes, they can be added to the Metric or MetricCalculation classes.

  2. New Models: For entirely new types of metrics or data, new dataclasses can be created following the existing pattern.

  3. Inheritance: The existing classes can be extended through inheritance if more specialized metric types are needed.

This metric-focused design ensures that Orbis can efficiently handle a wide range of metrics while remaining flexible for future additions and changes.