Migration from Informatica PowerCenter to ODI: accelerated with Sql2Odi
Streamlining ETL Migrations with Sql2Odi
I have previously blogged about Rittman Mead's Sql2Odi tool, which translates SQL statements (SELECT or WITH) into Oracle Data Integrator (ODI) Mappings. After years of use, we recently put Sql2Odi to the test during a large ETL migration project. Our goal was to convert Informatica PowerCenter Mappings, which often embed data join and transformation logic in SQL, to ODI Mappings.
The beauty of Sql2Odi lies in its ability to leverage existing SQL code, without having to re-implement all legacy ETL logic by hand. With minimal modifications, we fed query statements from PowerCenter into Sql2Odi and generated fully functional ODI Mappings.
Beyond Basic SQL: Handling Complex Queries
Sql2Odi goes beyond simple queries. It can handle even intricate SQL and generate fully working ODI Mappings:
- Complex SQL Support: Process intricate logic embedded within your SQL statements, including joins, subqueries, unions, and analytic functions.
- Target Table Definition: Specify the table where the data will be populated.
- Column Mapping Flexibility: Define how source data maps to target columns using names or positions.
- IKM, LKM, and CKM Configuration: Set up Integration, Lookup and Check Knowledge Modules (IKM, LKM, CKM) and configure their parameters.
- Scenario Generation: Request the creation of a Scenario for the generated ODI Mapping.
- ODI variables are supported inside strings.
Limitations
While Sql2Odi offers a lot of functionality, here are some limitations:
- Nested Query Support: Nested SQL queries are not currently supported. (Subqueries are supported.)
- No Query Lookup Handling: Lookups like
WHERE COUNTRY IN (SELECT COUNTRY_NAME FROM ...)
are not supported but usually can be re-written as joins. - Hierarchical/Pivot Query Limitations: Hierarchical and pivot queries are not supported. Some simpler hierarchical queries can be re-written as joins.
- Oracle (+) Join Notation: The Oracle (+) join syntax cannot be directly translated. (The (+) notation can be converted to ANSI joins with the latest versions of Oracle SQL Developer. ANSI joins are supported by the Sql2Odi tool.)
Sql2Odi in Action: let the Numbers Speak
In my previous blog posts, I emphasized Sql2Odi's ability to handle complex SQL. Now, I have the numbers to prove it:
- 92 SQL Statements Parsed: During the ETL migration project, a significant number of real-world SQL query statements were successfully processed, some of them very large and complex.
- Nearly Half a Megabyte of SQL Converted: These 92 queries translated to a combined size of over 448,220 characters, demonstrating the tool's ability to handle substantial amounts of code.
Parsing Efficiency and Fine-Tuning Opportunities
The SQL2Odi tool follows a two-step process: SQL parsing and ODI content generation. Parsing the 92 statements is remarkably fast, taking less than 7 seconds for our tool. Generating ODI Mappings takes slightly longer, with each Mapping requiring between half a second and a few seconds. (Mapping validation and Scenario generation by the Sql2Odi tool costs extra execution time but also adds value.) While the generated Mappings are fully functional and work correctly, an ODI developer may find opportunities to streamline them for conciseness, especially the Mappings generated from WITH statements.
Proven Success in Large-Scale ETL Migration
This successful project demonstrates the power of Sql2Odi in streamlining ETL migrations and reducing development time. The tool's reliability has been validated in a large-scale project, showcasing its capability to handle even large and complex SQL queries effectively.