Oracle Fusion Cloud ERP and Looker integration

What moves between them

Financial records flow from Oracle Fusion into Looker. ml-connector polls Oracle Fusion for new and modified invoices, payments, journal lines, and GL account balances on a schedule tied to your financial close calendar (every 5 to 15 minutes for near-real-time or daily for end-of-day batches). Those records are transformed into Looker's modeled schema and loaded into the warehouse. Looker scheduled plans then extract subsets (e.g., AP aging, GL account reconciliation, cash position) and deliver them to email inboxes, S3 buckets, or SFTP destinations on a schedule you define. Vendor and GL account master data can align in both directions so Looker reports aggregate correctly by supplier and cost center.

How ml-connector handles it

ml-connector stores Oracle Fusion pod URL and OCI Identity Domain credentials encrypted and obtains Bearer tokens via OAuth 2.0 client credentials grant, refreshing tokens when they expire (approximately 1 hour). On the Looker side, it exchanges client_id and client_secret for a bearer token and refreshes on each call expiry. Because Oracle Fusion publishes no direct webhooks, ml-connector polls the /invoices, /payments, /journalLines endpoints with filtering on LastUpdateDate>={ISO8601_timestamp} and respects OData query parameters (limit, offset) to paginate large result sets. Looker's scheduled plans do not natively accept direct data pushes; instead, ml-connector coordinates with pre-built Looker queries or data actions that populate warehouse tables, and Looker scheduled plans then distribute the modeled output. Rate limits are not documented for Oracle Fusion, so ml-connector applies conservative backoff on retries. Every poll and transformation carries a full audit trail and can be replayed if a downstream Looker query or scheduled plan delivery fails.

A real-world example

A mid-sized professional services firm runs Oracle Fusion Cloud ERP for accounting, procurement, and project management. Finance teams use Looker for daily cash position, AP aging, and project profitability dashboards but had to export Oracle Fusion reports manually each morning and load them into a staging table before Looker queries could refresh. With Oracle Fusion and Looker connected, invoices, payments, and journal lines flow into Looker's warehouse every 15 minutes. Scheduled plans automatically deliver a daily cash position report to the CFO inbox at 7 AM and an AP aging roll-up to the procurement team at noon. Month-end close analysts query live GL balances in Looker without waiting for manual extracts.

What you can do

• •Poll Oracle Fusion for invoices, payments, journal lines, and GL account balances on a schedule tied to your financial close calendar.
• •Transform Oracle Fusion financial records into Looker's modeled schema and load them into the warehouse for querying and dashboarding.
• •Align vendor master data, GL accounts, and cost centers between Oracle Fusion and Looker so reports aggregate correctly by dimension.
• •Authenticate both systems via OAuth 2.0 client credentials, refresh tokens on expiry, and securely store all credentials encrypted.
• •Schedule automated delivery of financial reports from Looker to email, S3, or SFTP destinations on a cadence you control.

Questions

Which direction does financial data flow between Oracle Fusion and Looker?

Data flows from Oracle Fusion into Looker only. ml-connector polls Oracle Fusion invoices, payments, journal lines, and GL accounts and loads them into Looker's warehouse for modeling and analysis. Looker does not write back to Oracle Fusion financial records; it is a read-only analytics layer. Master data such as vendors and GL accounts can be validated in both directions to ensure Looker reports slice correctly.

How does ml-connector handle Oracle Fusion's lack of webhooks and the 1-hour token expiry on both sides?

Because Oracle Fusion publishes no outbound webhooks, ml-connector polls the REST API on a schedule (every 5 to 15 minutes for real-time or daily for batch) using OData query filters on LastUpdateDate to find only new and changed records. For token expiry, ml-connector tracks Looker's 1-hour bearer token and refreshes it by re-authenticating with client_id and client_secret. Oracle Fusion tokens follow the same pattern, so both systems stay authenticated without manual intervention.

What happens if a Looker scheduled plan fails or a report delivery is delayed?

ml-connector tracks the full audit trail of every poll from Oracle Fusion, transformation, and Looker scheduled plan trigger. If a downstream Looker query or email delivery fails, the original financial records are retained in the audit log and the scheduled plan can be re-run or manually corrected without losing data. Retries and backoff logic handle transient failures on both the Oracle Fusion polling side and the Looker API side.