Skip to main content

Data modes and datasets: live, audit and sim

Every result says which data it read, so live, audit and simulated figures are never mixed. Telemetry lands in one of three datasets, and the Platform labels each view by what it shows: Live, Audit, Sim, Scenario or Demo fleet.

The demo fleet is kept apart by its own satellites and antennas: its simulated telemetry names only them. The demo's workflows price only the demo's satellites, its planted rain reaches only the demo's antennas, and the Platform marks its rows Demo and leaves it out of the counts on Integrations. To keep it wholly apart, build your own fleet in a new workspace, or remove the demo.

The three datasets​

DatasetWhat it holdsHow it gets there
liveWhat your links measured, as it happenedPOST /telemetry or the fleet agent. A record that names no dataset is live.
auditThe contacts you flew, from a pass logPOST /telemetry/pass-logs writes every row into audit
simSimulated telemetryA telemetry record sent with dataset: "sim"

A pass log never mixes with live telemetry: uploading one for an audit leaves your live series as they were, and uploading the same log again rewrites the same records rather than counting them twice.

Which data a result read​

LabelMeansWhere you see it
LiveA run over your network config and live telemetryPasses: Upcoming, and Past with Live chosen
AuditA run that set past passes against an uploaded pass log (dataset: "audit" on the run)Past on Passes, with Audit chosen
SimA run that read the sim datasetWherever such a run is shown
ScenarioA run against a what-if (scenario_id on the run). It delivers nothing.Scenarios on Forecasts, framed in its own color
Demo fleetYour network config is the demo fleet, not yoursFleet, Passes, Forecasts and Automations, in a frame labeled Demo fleet

A run's dataset matters only to a workflow with a ledger step, such as audit-ledger: it names the dataset the ledger reads logged contacts from. Send it for those workflows only; see Running and reading a workflow.

Runs that deliver nothing​

Two kinds of run compute and store their results but send nothing to your mission control system (MCS) or destinations:

  • A test run (deliver: false), such as Price the next 48 hours on Passes, or Test in the builder.
  • A scenario run, whatever it sends for deliver, so a what-if never reaches an operator as a forecast.

See Runs and test runs.

Open data​

The demo fleet is also published as an open dataset: the same network config and telemetry the API loads, to study offline. See Hugging Face.