OBJECT MODEL
TRAC is built around a structural metadata model which describes, catalogues and controls everything on the plartform. Each object type has its own metadata structure which is built up from the most foundational objects.
DATA
Data is stored using an append-only data model so a complete version history is retained. Data items and is represented within the platform using schema’s. TRAC supports single parameters, structs, and tables.
MODEL
Models are not deployed in the physical sense. The code is stored in a repository and fetched dynamically at runtime. Models are represented via a 'model schema' – the set of schemas that fully describe the models input, outputs and parameters. TRAC knows that model A requires inputs X and Y and generates output Z - it just doesn't know why.
FLOW
A flow is the blueprint of a complex calculation that can be run on TRAC involving multiple models, where the outputs of one model become the inputs to the next. They are built using TRAC's FlowBuilder tool and exist only in schema – or graph – form. They represent the virtual deployment of your models into an execution process.
JOB
To run a job you select a flow, populate it with model and data objects and provide any numerical parameter values via the UI. The data records and code are then fetched, and the job is executed. The schema for the entire job is stored as metadata, so every calculation can be described automatically, without needing manual documentation.
Explore Virtual Deployments
