IBM InfoSphere DataStage



IBM InfoSphere DataStage is an ETL tool and part of the IBM Information Platforms Solutions suite and IBM InfoSphere. It uses a graphical notation to construct data integration solutions and is available in various versions such as the Server Edition and the Enterprise Edition.

A data extraction and transformation program for Windows NT/2000 servers that is used to pull data from legacy databases, flat files and relational databases and convert them into data marts and data warehouses. Formerly a product from Ascential Software Corporation, which IBM acquired in 2005, DataStage became a core component of the IBM WebSphere Data Integration suite.

DataStage originated at VMark[1], a spin off from Prime Computers that developed two notable products: UniVerse database and the DataStage ETL tool.

The first VMark ETL prototype was built by Lee Scheffler in the first half of 1996[1].

Peter Weyman was VMark VP of Strategy and identified the ETL market as an opportunity. He appointed Lee Scheffler as the architect and conceived the product brand name "Stage" to signify modularity and component-orientation[2].

This tag was used to name DataStage and subsequently used in related products QualityStage, ProfileStage, MetaStage and AuditStage.

Lee Scheffler presented the DataStage product overview to the board of VMark in June 1996 and it was approved for development.

The product was in alpha testing in October, beta testing in November and was generally available in January 1997.

VMark acquired UniData in October 1997 and renamed itself to Ardent Software[3]. In 1999 Ardent Software was acquired by Informix[4] the database software vendor.

In April 2001 IBM acquired Informix and took just the database business leaving the data integration tools to be spun off as an independent software company called Ascential Software[5].

In November 2001, Ascential Software Corp. of Westboro, Mass. acquired privately held Torrent Systems Inc. of Cambridge, Mass. for $46 million in cash.

Ascential announced a commitment to integrate Orchestrate's parallel processing capabilities directly into the DataStageXE platform. [6].

In March 2005 IBM acquired Ascential Software[7] and made DataStage part of the WebSphere family as WebSphere DataStage.

In 2006 the product was released as part of the IBM Information Server under the Information Management family but was still known as WebSphere DataStage.

In 2008 the suite was renamed to InfoSphere Information Server and the product was renamed to InfoSphere DataStage[8].

•Enterprise Edition: a name give to the version of DataStage that had a parallel processing architecture and parallel ETL jobs.

•Server Edition: the name of the original version of DataStage representing Server Jobs. Early DataStage versions only contained Server Jobs. DataStage 5 added Sequence Jobs and DataStage 6 added Parallel Jobs via Enterprise Edition.

•MVS Edition: mainframe jobs, developed on a Windows or Unix/Linux platform and transferred to the mainframe as compiled mainframe jobs.

•DataStage for PeopleSoft: a server edition with prebuilt PeopleSoft EPM jobs under an OEM arragement with PeopleSoft and Oracle Corporation.

•DataStage TX: for processing complex transactions and messages, formerly known as Mercator.

•DataStage SOA: Real Time Integration pack can turn server or parallel jobs into SOA services.



Monday, November 2, 2009

T-ETL architecture using WebSphere DataStage and WebSphere Federation Server

This article focuses on the benefits of using the combination of WebSphere DataStage and WebSphere Federation Server to perform data consolidation. It promotes the use of WebSphere Federation Server as a data pre-processor for WebSphere DataStage, essentially performing initial transformation either before, or as the data is extracted from one or more sources. The T-ETL architecture proposed uses federation to join, aggregate, and filter data before it enters WebSphere DataStage, with WebSphere DataStage using its parallel engine to perform more complex transformations and the maintenance of the target.
The proposed architecture is illustrated in
Figure 1:









The architecture draws on the strengths of both products, producing a flexible and highly efficient solution for data consolidation; WebSphere Federation Server for its joining and SQL processing capabilities, and WebSphere DataStage for its parallel data flow and powerful transformation logic. The WebSphere Federation Server cost-based optimizer also allows the T- ETL architecture to dynamically react to changes in data volumes and patterns, without the need to modify the job.

T-ETL is not a new concept, and many ETL jobs may already employ some form of transformation while extracting the data -- such as filtering and aggregating data, or performing a join between two source tables, which reside on the same source database. However, the restriction that the source objects must exist on the same data source has severely limited the scope of T-ETL solutions to date. WebSphere Federation Server removes this limitation and extends this initial transformation stage to heterogeneous data sources that are supported by WebSphere Federation Server.

For example, WebSphere Federation Server allows T-ETL when the source data is an Oracle table, a Teradata table, and a flat file. In addition to extending the scope of the initial transformation stage, federation is also able to improve on the efficiency of the stage since at its core, it is a relational database engine with more than 30 years investment in efficiently filtering and joining data sets.
Before Next

Labels:

0 comments:

Post a Comment

Subscribe to: Post Comments (Atom)