ATX AnalyticsTable connects to Oracle Database in Thin mode, so there is no Instant Client to roll out, and works on the results locally. It is built for the core-system reports that cannot go to a cloud: the query runs in Oracle, the rows land in a DuckDB database on your machine, and national IDs, surnames and account numbers are masked before any extract leaves it.
The Oracle Database connection form in ATX AnalyticsTable 1.7, filled in with the example values below. Secrets go into the encrypted vault, never into the project.
A read-only reporting user (as a DBA): CREATE USER report_ro IDENTIFIED BY "<a long password>"; GRANT CREATE SESSION TO report_ro; GRANT SELECT ON core.accounts TO report_ro; GRANT SELECT ON core.customers TO report_ro; — SYS / SYSDBA is not offered in the form — use an ordinary user with SELECT on what it reads.
Reporting query — national IDs and surnames are reduced inside Oracle: SELECT a.branch_code, TRUNC(a.opened_on, 'MM') AS month, RAWTOHEX(STANDARD_HASH(c.national_id, 'SHA256')) AS customer_key, SUBSTR(c.last_name, 1, 1) || '***' AS last_name_masked, SUM(a.balance) AS balance FROM core.accounts a JOIN core.customers c ON c.customer_id = a.customer_id WHERE a.opened_on >= ADD_MONTHS(TRUNC(SYSDATE, 'MM'), -12) GROUP BY a.branch_code, TRUNC(a.opened_on, 'MM'), RAWTOHEX(STANDARD_HASH(c.national_id, 'SHA256')), SUBSTR(c.last_name, 1, 1) || '***' ORDER BY month FETCH FIRST 50000 ROWS ONLY — STANDARD_HASH needs Oracle 12c or later. No trailing semicolon: the statement is sent as one call.
Then reshape locally — a Run SQL step on the project's DuckDB: -- "input" is the table the previous pipeline step produced (the import above). -- This runs on your machine, in the project's DuckDB, not on the database. SELECT region, SUM(revenue) AS revenue, COUNT(DISTINCT customer_key) AS customers FROM input GROUP BY region ORDER BY revenue DESC; — Column names follow the query above; on the local engine the pipeline's current table is registered as input before the script runs.
Oracle Data Redaction and Virtual Private Database policies apply to the database user that runs the query. With a password, wallet or external login everyone shares that one user. A token connection with a token scope can run as the signed-in person through on-behalf-of, and the person's name is written into CLIENT_IDENTIFIER, so V$SESSION and fine-grained audit record who asked.
Oracle usually runs inside your network already, so “local” here means no cloud at any point: the client, the copy of the data, the masking and the output are on your machine or your server. After the import, reports and pipelines over the local copy run without a database connection; the licence validates offline.
No. The connector uses Thin mode; host, port and service name — or a connect string, a tnsnames alias or a full descriptor — are enough.
No. Masking in the Mask PII tool and the pipeline step is one-way and keeps no map back to the originals. Hash and Synthetic are stable — the same customer gets the same token every run — because they are keyed with the project's random key.
With a token connection and on-behalf-of, the signed-in person's name goes into CLIENT_IDENTIFIER, so V$SESSION and fine-grained audit show who asked. With a shared user, Oracle sees that user.
This site needs JavaScript for the full experience. Every page above is reachable without it.