Files (SFTP or S3)

You export files and deliver them on a schedule to SFTP or S3. Athia provides the destination, the format specification and the validation.

📘

If your payments data already sits in a warehouse or data lake, read Warehouse to Warehouse first. Athia reads Snowflake, Databricks, BigQuery and Redshift directly from a grant your admin defines, which is usually less work than building an export: the joins are already done, and history and the recurrent feed come from the same grant with no separate backfill to arrange. Files are the right answer when there is no warehouse, when payments are not modelled in it, or when policy rules a direct grant out.

  • Who owns it on your side: your data team. They export from the systems of record and schedule the delivery.
  • What Athia gives you: the SFTP or S3 destination and credentials, the field specification (see Athia Data Dictionary), and a validation report on every delivery listing rejected rows, missing required fields and type mismatches, checked against the mapping agreed for your account rather than against fixed column names, so you can correct the export before the next run.
  • Cadence: daily is typical for the recurrent feed. Match it to how often your source systems close.
  • Formats: Apache Parquet is preferred. Gzip CSV, JSON Lines, or direct table and view access also work.
  • File granularity: one file per source per month, around one gigabyte per file, split if larger.
  • A one-time effort: the historical export is done once. Once the live connection runs, no ongoing manual updates are needed.
Backfill and validation loop
FILE DATA DUMPSFrom first export to a live feed 1Export history attempt level, not orders 2Deliver to SFTP or S3 3Athia validates against your agreed mapping Validation report Rejected rows Missing required fields Type mismatches Correct the export and re-deliver clean Switch the same export to a schedule Recurrent feed is live The historical export is a one-time effort. Once the live connection runs, no ongoing manual updates are needed.

The backfill and the recurrent feed use the same path and the same specification, so the work is done once. Athia validates every delivery and returns a report naming rejected rows, missing required fields and type mismatches. Correct the export rather than the file, or the next scheduled run repeats the problem.

Your deliveryCorrection loopLive

The backfill and the recurrent feed use the same path and the same specification.

Delivering over SFTP

SFTP is the right choice for bulk history and for feeds that do not need to arrive within seconds.

Getting access

Ask your Athia team to open an SFTP account. You exchange public keys, and they return the host, the username, and the upload and receipt directories for your account. There is no password.

Naming your files

Start every filename with the type of data it holds. Athia routes on that prefix, so a misnamed file is a misrouted file.

  • transactions_2026-05-18_01.parquet
  • costs_2026-05-18_01.parquet
  • subscriptions_2026-05-18_01.parquet

Everything after the prefix is yours to use: a timestamp, a source system, a part number.

Uploading, and reading the receipt

  1. Connect with the host and username you were issued.
  2. Drop your files in the upload directory.
  3. Read the receipt in the receipts directory on the same server.

Athia writes a receipt for every file it processes, naming rejected rows, missing required fields and type mismatches. Correct the export rather than the file, or the next scheduled run repeats the problem.


Did this page help you?