Oracle announced this week updates of its Oracle Data Integrator and GoldenGate data-integration products that do much more than support the Oracle 12c database released this summer. Oracle Data Integrator 12c introduces a complete makeover of the job-building experience, while real-time oriented GoldenGate 12c introduces performance gains.
Oracle Data Integrator (ODI) is different from most data-integration products in that it's an extract, load, transform system (not extract, transform, load) that relies on the database to handle transformation rather than relying on middle-tier ETL servers. The ODI 12c upgrade retains the power and cost advantages of this approach, according to Oracle, but it also adds the flexibility and familiarity of visual, flow-based job design used by most conventional ETL tools.
"With ODI, you had to learn how the product wanted you to do the mapping," said Brad Adelberg, Oracle's VP of development, in an interview with InformationWeek. "With traditional ETL tools you have a pallet of operators -- read from source, filter, join, sort, aggregate -- and you can arrange them any way you like."
The flow-based tools are very flexible, but you can get buried in the weeds of job design, and job flows are brittle if there are changes to bits and pieces of the flow, according to Adelberg. ODI 12c gives users the best of both worlds, he said, combining both flexible, visual design and the high-level abstraction of ODI, which takes care of many of the changing details of job execution.
[ Want more on Oracle's new database? Read The Truth About Oracle's Virtualized 12c Database. ]
ODI's new approach will make integration work more intuitive for the many practitioners who are used to flow-based products like Informatica Power Center and IBM DataStage, Adelberg said. Yet everything you built in 11g will still work in ODI 12c, and you still use the current development approach.
GoldenGate is Oracle's change-data-capture (CDC) product for real-time data integration, and the big advance in GoldenGate 12c -- in addition to support for the multitenant 12c database -- is integrated delivery. It's an approach in which changes from a source database are applied directly within an Oracle target database, extending a data-reading advance that was introduced in the last release of the product.
"Rather than reading redo logs, GoldenGate 11 brought the advance of capturing changed data directly from the database," Adelberg explained. "In GoldenGate 12c, we're doing the same thing on the delivery side when an Oracle database is the target of replication."
So when an Oracle database is either the source or the target of change-data replication, this feature delivers higher performance as well as the ability to work with compressed and encrypted data. Previously it was impossible to change such data without going through the step of decompression or decryption.
ODI and GoldenGate serve different roles. ODI is Oracle's standard, batch-transformation tool, used primarily in data warehousing scenarios. GoldenGate is a real-time data-replication product, but it has limited transformation capabilities. Another advance in ODI 12c lets companies tap GoldenGate's change-data-capture capabilities from within the ODI interface so customers can blend batch and real-time data integration. ODI alone can't keep up when applications call for up-to-date information within seconds or a few minutes.
"ODI Studio is now capable of configuring, orchestrating, starting and stopping GoldenGate processes, but you don't have to become a GoldenGate expert," Adelberg said.
Before Oracle acquired GoldenGate in 2009, the company worked with a who's who list of databases and applications. Oracle vowed to maintain support for heterogeneous environments, and in 12c, Adelberg said, Oracle is delivering up-to-date support for MySQL, Sybase, IBM DB2, Teradata, Microsoft SQL Server and other databases and myriad packaged applications. Oracle has also added integrations with Oracle Coherence, whereby GoldenGate can duplicate changed data within the Coherence distributed object cache.
GoldenGate can't do in-database capture and delivery with third-party databases as it can with Oracle database (it has to rely on the old, log-reading approach to CDC). But if Oracle is either the source or target database used in conjunction with other products, customers can get at least half of the performance boost of the latest GoldenGate release.