Metadata management

Horizonte Operator Can Locate a call Column estadobut you need to know what it means, who's responding for her, how she's been changed and what reports or agents depend on her contents. The catalogue is useful when it helps to solve these questions with evidence.

Key concepts

The Business metadata describe meaning, goal, owner and rules. The technical describe schemes, columns, types and relationships. The operations record execution, dates, results and update. The three complement them: knowing the type of a column doesn't explain how to use it in a decision.

A glossary He agrees with terms of his business. A catalogue allows to organize and discover assets. The lineage links sources, transformations and results and impact helps to identify what could be affected by a change. A description suggested by AI is a draft until a owner values its meaning.

The estado_contrato column requires three different contextual classes. The company defines "valid contract," owner and use. The technician identifies type, column and rule R-07. The operative records execution E-308, an update and monitoring result. The three links to interpret the value and check how it was produced without confusing description with evidence from execution. ↗

Linkage, Impact and Responsibility

The steward relates "valid contract" with the columns and rules that implement it. The engineer records transformations and executions. The owner approves the definition. When a source changes, lineage helps to track its path and impact analysis identifies consumers who have to review their assumptions.

The range matters: a graph showing a table and a notebook doesn't prove that you know all export, spreadsheets or outside queries. The evidence should declare cover, granularity and period captured. The units known outside the tool are documented as part of the process.

C-17 has an active contract, passes rule R-07 and contributes one distinct customer to the list used by Customer Service; pending C-28 is excluded. To establish provenance, trace that result back to the contract and rule. To assess impact, consider a specific change: allowing pending contracts includes C-28, changing the result from 1 to 2 customers and the Customer Service list. A dashed line connects the metric to its exported spreadsheet, whose dependency must be documented outside captured scope. The owner approves the definition, the steward links the term to R-07, and the engineer preserves its version and run. ↗

Horizonte registers owner from the definition, owner technical asset and steward that checks the context. Those roles can fall to few but they have to be explicit. The catalogue should not become an asset list without decision or maintenance.

Catalogs and Linkage in Oracle

Use Master Catalog from Oracle AI Data Platform To organize catalogs, schemes and assets and manage their permissions. In Data Studio from Oracle Autonomous AI Database, explores connected sources and sets up catalogs using an interface or DBMS_CATALOG. To manage tables Iceberg from supported engines, configure Oracle AI Data Catalog, his REST service, his credentials and his storage access. At each connection, it identifies which metadata are updated, who keeps the description and with which permissions the asset is consulted.

The rows compare three fields: Master Catalog from Oracle AI Data Platform for catalogs, schemes and assets. Oracle Autonomous AI Database's dedicated catalogue for supported sources and Oracle AI Data Catalog with Iceberg's REST interface. Check objects, interfaces, compatibility and permissions for each integration. Do not have synchronization or shared access by the fact that all three are catalogs. ↗

To analyze a change in Oracle AI Data Platform, open the asset at Master Catalog and select Linage. Travel sources and consumers or expand columns to study their derivations. The view shows the last capture of each process. preserves the versions and performance of the period you have to conduct. Data Studio from Oracle Autonomous AI Database uses Linage e Impact in the details of the asset to review the relations of captured flows and loads.

It links the term "valid contract" with the tables, columns and rules that implement it. At Oracle Data Transforms, matters, edits and spreads notations To preserve that context at destination tables. The steward checks the description with the owner, records its approval and updates the affected assets as they change their meaning. It applies the same review to the descriptions suggested by AI to maintain coherence between glossary and implementation.

Explanated example

The Contracts team wants to change the meaning of estado=1earlier referred to as "valid" Now would have been "tipping up." The change conserves type and name, so an exclusively technical check could accept it.

The steward looks at the approved meaning and goes through dependencies: calculation of active customers, monthly report and context used by AI. The owner decides to separate the states rather than quietly extend the meaning. Teams update rules and descriptions and record an effective date.

The proposal keeps state = 1 and the same type of column but extends the meaning from "valid" to "valid + outstanding." That change can alter active customers, reports and responses from Customer Care without changing the scheme. The owner should solve the definition, update rules and descriptions and register approval and effective date before accepting the new meaning. ↗

The evidence brings together earlier and new definition, revised consumers, approval and reconciliation results. A measurable improvement is to reduce critical assets without owner or approved terms without link to its implementation, without confusing quantity of quality descriptions with government.

Common

• "A catalogue automatically creates the glossary." Uncover columns doesn't agree with corporate meaning.
• "All lineage is historic and complete." Coverage, granularity and period caught have to be declared.
• "The AI description is already approved." The generation does not replace the validation of steward and owner.

Decision exercise

Audit wants to rebuild the transformation that was used three months ago. The lineage view of Oracle AI Data Platform shows the last capture and the notebook changed several times. Decide what evidence of that execution should gather Horizonte to explain the historic result.