Data freshness

Pricing a freight chain makes zero live external calls. Every input the cost engine needs — fuel prices, toll rates, route geometry, vehicle positions — is pre-synced into the database by background jobs. So pricing a chain is a pure read plus math: fast, deterministic, and dependent on nothing outside the database at the moment of pricing.

Why DB-only

Live lookups would make pricing slow, non-deterministic, and fragile — a chain's profit would depend on a third-party API being up and fast at the instant a dispatcher (or fOS) asks. Instead, TOVR separates gathering the data from using it: background jobs gather it on their own schedule; the cost engine only ever reads what's already there. The result is that a chain re-prices identically whether it's the first request or the thousandth, online or not.

What gets synced, and how often

Background jobs keep each input current on its own cadence:

Toll and route-geometry data are likewise cached rather than fetched per request. The exact schedules are an operational detail; what matters conceptually is that the data is already there when the engine runs.

Freshness becomes cost confidence

Pre-synced data is occasionally incomplete — a price not yet synced for a country, a route not yet cached. When an input is missing, the engine degrades gracefully rather than blocking: it falls back to a default and lowers the chain's cost confidence rather than failing the price. Cost confidence is broadly:

• calculated — all inputs present; full confidence.
• estimated — some inputs came from defaults or benchmarks.
• partial — meaningful gaps in the data behind the price.

Confidence lets a dispatcher (and fOS) weight how much to trust a P&L, and it recovers automatically on the next sync as the caches refill. See Cost engine › cost confidence.