Model Manager: service that collects and persists all of the metadata, execution, and test results about a model. This service is backed by a MongoDB for persistence, and provides the production Model Catalog as well as all of the Model Details for each model.
MLC (model life cycle) Manager: service that orchestrates and manages all of the processes in a model’s entire life cycle, from registration, to promotion, to production, to monitoring, and eventual retirement.
ModelOp Runtime: language and framework-agnostic model execution runtime, allowing for models of all types to be executed in batch, streaming, or REST mode.
Gateway: service that provides secure entry into all other core ModelOp services.
Registry: registration service for all other micro services within ModelOp Center, enabling automatic registration of new instances of core ModelOp Center services as well as on-going monitoring and management of each service.
Reporting: flexible service to allow for reporting from MLC Manager as well as other internal and external runtimes and services.
Document: service that allows for automatically generating documents using model metadata, test results, and other model information from document templates.
Optional micro services:
Model Factory Plugins: set of plugins into common model factories to enable easy registration and management of models that are created within those model factories. Jupyter and RStudio are the main existing supported plugin.
Mode Execution (Runtime) Services: services to integrate with various model execution runtimes, such as Spark, SageMaker, etc.
Logstash: output from ModelOp services and instructions on how to configure logstash to write to your preferred location.
Click Here to learn about the Terminology of ModelOp Center
Click Here to Get Oriented with ModelOp Center’s Command Center
Click into the series of articles below or to learn how to Deploy, Monitor, or Govern models in ModelOp Center:
Next Article: ModelOp Center Terminology >