Timber
Overview
Timber is the CaraML component responsible for log collection, model monitoring, and dataset creation. It consists of several components:
Dataset Service
Dataset service is the orchestrator within Timber ecosystem. It manages deployment of log writer and observation service.
Observation Service
Observation Service provides an interface for reporting observations from Client-owned systems back to CaraML platform. The logged observations will be used for evaluating the effectiveness of a given prediction instance.
Log Writer
Log writer is the component responsible for collecting prediction log and router log and store it in the persistent storage to be used for downstream use case such as dataset generation or model monitoring.
Contributing
Timber is still under active development. Please have a look at our contributing and development guides if you want to contribute to the project:
Project Structure
├── common # Common code to be reused both in Dataset service and Observation service
├── dataset-service # Dataset service implementation.
├── images # Docker images
│ └── fluentd # FluentD docker image for logwriter and observation service
| └── observation-service # Dockerfile for building Observation service
├── infra # Infrastructure setup for testing and deployment.
├── logwriter # Logwriter implementation.
│ └── fluent-plugin-upi-logs # FluentD plugin for parsing Universal Prediction Interface protobuf.
├── observation-service # Observation service implementation.
├── scripts # Miscellaneous scripts.
| └── vertagen # Helper for version string generation.
├── tests # Integration and end to end test suites.
├── CONTRIBUTING.md # Contributing guide.
├── LICENSE
├── Makefile
├── README.md