Migrate your Timefold data

Timefold's core artifact — a Model defines the planning domain (entity types, variable fields, and constraint rules) for a specific problem type such as Field Service Routing or Employee Shift Scheduling. Models are selectable from Timefold's pre-built library and are referenced by name/version in API calls. We migrate Model configurations by serializing the domain class structure and constraint definitions to the target environment.

Planning Entities are the classes whose instances change during solving — for example, Visit, Shift, Job, or Route. Each entity has one or more planning variable fields that the solver populates. We map these from the source system (e.g., a work-order list) into Timefold-compatible POJOs with Jackson annotations for JSON serialization.

Problem Facts are immutable data used to evaluate constraints — technician skills, vehicle capacities, time windows, skill requirements, employee availability. We extract these from the source system and feed them as read-only input to Timefold. They remain unchanged by the solver.

Planning Variables are the fields on entities that Timefold assigns during optimization (e.g., which technician is assigned to which visit, or what time a shift starts). Because variable types and naming vary by model, we perform schema mapping to align source field names and value formats with Timefold's expected variable structure.

A Planning Solution wraps all entities, facts, and the score (hard/medium/soft constraint results) into a single output object returned by the solver. We extract the solved assignments from this object and write them back to the destination system as records.

Timefold Platform supports multi-tenant organization. A tenant holds one or more models and members with role-based visibility. We migrate tenant configuration (names, member assignments, roles) and ensure the target environment recreates the same organizational boundaries.

Members are users within a tenant. Roles control access to models and API endpoints. We map member records and role assignments to the destination platform's equivalent user/permission model.

The solver returns a score with three tiers: hard (infeasibility violations), medium (resource scarcity), and soft (optimization objectives like travel time or fairness). Scores are exported as metadata alongside the solved assignments so downstream systems can assess solution quality.

In the Employee Shift Scheduling model, Shifts are planning entities that get assigned to Employees. We migrate historical shift data as problem facts and future shift assignments as the solver's output planning solution.

In Field Service Routing, Visits represent customer appointments; Routes are the sequence of visits assigned to a vehicle or technician. We extract visit lists with time windows and skill requirements, run them through Timefold, and import the optimized route sequences.

Vehicles are problem facts in the routing models with properties such as capacity, start/end locations, and operating hours. We map vehicle fleet data from the source system into Timefold's Vehicle class schema.

Timefold allows custom constraint definitions (DRL rules or Constraint Streams in Java). When migrating, we assess whether custom constraints are in use and translate them to the equivalent format for the destination solver, or flag where destination constraint expressiveness differs.

Timefold provides built-in demo datasets (e.g., BASIC) for testing the Field Service Routing model. These are reference data, not customer data, and are excluded from migration scope.

Timefold Platform supports incremental re-optimization via patches — submitting an emergency visit, a technician illness, or a no-show without re-running the full problem. Patches are an API-level concept; we map these to the equivalent real-time update mechanism in the destination system if supported.