A manual export today.
A governed Azure medallion with a full audit trail.
Trainline's numbers are made in AWS Spark and posted into Dynamics 365 through a hand-run Tableau export. Pigment draws from two routes that nobody reconciles. This is an interactive look at where the estate is now and the two-phase Azure target that replaces the manual path and makes every figure traceable back to raw.
Manual in, compressed out, nothing traceable
Modelled output leaves AWS as a Tableau CSV and is uploaded into the ledger by hand, compressed 3,500:1 to 5,000:1 with no key to trace it back. The outbound route to planning still runs a full historic refresh to 2018 and has broken under volume.
Athena ODBC and Synapse Link into Bronze
An Azure VM running the Amazon Athena ODBC driver pulls AWS data across; the D365 team publishes Dataverse through Azure Synapse Link. Both land raw into Bronze on an ADLS Gen2 account with hierarchical namespace enabled.
Silver, Gold, then Pigment and D365
Silver conforms both sources; Gold applies the aggregation grain, measures and batch reference key agreed with finance. Gold serves Pigment and posts back into D365, so both sides come from one place.
A full audit trail, both directions
Bronze keeps every record exactly as it arrived from AWS and from Dynamics 365. That means any transformed figure can be retraced back to the raw data behind it, and any raw record can be followed forward to where it landed. Processing can run on Microsoft Fabric, Azure Synapse, or Daxia - our own web app deployed in your tenant at no licence cost.