Glossary

These section provides definitions for terms which have a specific meaning for TRAC.

Asset

A data or model object, plus the external asset (code or data records) to which the metadata object refers

Business segment

A user-defined classifier which forms part of the user tags

Data

Depending on context either a) collections of files or records, or b) a metadata object of the type ‘data’ which refers to and describes those files and records

DataStore

See Internal storage

Execution service

An external compute service which TRAC uses to orchestrate RunModel and RunFlow jobs

ExportData job

A job which TRAC orchestrates to copy data from an internal storage location to an external storage location (not available in COMMUNITY TRAC)

External storage

A storage location to which TRAC has non-controlling access, from which data can be fetched via an ImportData job or to which data can be placed using an ExportData job (not available in COMMUNITY TRAC)

Flow

A metadata object of the type ‘flow’. Flows are blueprints for complex calculations which involve more than one model represented as a graph where the nodes are models and the edges are datasets

Governance report

A pdf copy of an object summary which is downloaded for use in offline governance processes

ImportData job

A job which TRAC orchestrates to copy data from an external storage location to an internal storage location (not available in COMMUNITY TRAC)

ImportModel job

A job TRAC runs when a model is loaded from a repository. The job creates a metadata object to represent and reference the code in the repository

Internal storage

A data storage location to which TRAC alone has write access, which is used to store imported, uploaded or generated data

Job

Depending on the context either a) a metadata object of the type job, which refers to and describes a process which TRAC has initiated, or the b) process itself. The five job types are; ImportModel, ImportData, RunModel, RunFlow and ExportData

Metadata model

The semantic data model which TRAC uses to catalogue, describe and control assets, resources and jobs. The model consists of two layers; objects and tags

Model

Depending on the context either a) a block of code in a repository that has been exposed to TRAC via the model upload process, or b) a metadata object of the type ‘model’ which refers to and describes said block of code

Object

A single record in the metadata store. The main object types; data, model, schema, file, flow and job

Object summary

The page in the TRAC UI which summarises a single object, including it’s tags, schema, version history, and any files which have been attached to it

Object tag

A group of metadata tags which TRAC automatically assigns to an object based on its type - e.g. model objects have tags for the model language and repository from which it was imported

Repository

An external code repository from which TRAC can import models via an ImportModel job. The repository will store the model code and provide it when needed for a RunModel or RunFlow job

Resource

Any technical service external to the platform which TRAC uses to orchestrate jobs. Resource types include; internal and external storage locations, repositories and execution services

RunFlow job

A job in which TRAC uses an execution service to orchestrate a calculation represented by a flow, using a specific set of models, data inputs and parameter values

RunModel job

A job in which TRAC uses and execution service to execute a single model using a specific set of data inputs and parameter values

System tag

A group of metadata tags which TRAC automatically assigns to every object - e.g. object type and creation timestamp

Traceable action

An action that a user performs on the platform which is recorded in the metadata model. Object creation and tag modification are the two types of traceable action

Upload

To make an asset available to TRAC by loading it directly to the UI rather than using an import job. Data, files and schemas can be uploaded

User tag

A group of metadata tags which are assigned to an object by a user for descriptive purposes and to aid object discovery - e.g. name, description, key and business segment