Inefficient Data Integration

In this Issue:

More vendors will provide architecture and "data services"

The aching need for more efficient data integration may soon find some relief. According to Philip Russom of Forrester Research, "data services" are on the rise. "Although data services are a bleeding-edge phenomenon at present, several vendors support them and a few user organizations have put them into production," says Russom.

Data integration is too often executed with point-to-point solutions, Russom points out. Hub-and-spoke designs historically prevail over point-to-point because they scale so much more economically. Data services, or a service-oriented architecture for data integration, can provide a hub that bestows the following benefits: Cost reduction by lessening the number of system interfaces you need to build and administer, better use of resources through reduced employee training and object reuse, system flexibility for faster accommodation of business changes, and enterprise visibility — essential to reducing risk exposure from government regulations.

What vendors support data services? With version 7 of DataStage, Ascential introduced Real Time Integration Services last summer. "Despite the focus on 'real time,' this is a data services layer," Russom contends. "RTI services is proving useful as a way to increase interoperability between products of Ascential's [heavily acquired] suite."

Also, Avaki recently released version 5 of its data grid product. "A few of Forrester's user clients use Avaki, and they corroborate that data services are an important component in virtualizing data resources, so they can be freely allocated and provisioned," Russom adds.

Russom cites most enterprise information integration (EII) products and embedded reporting as types of data services, as well.

When it comes to ease of data service creation, Russom holds out as the epitome Informatica's PowerCenter 7 design environment. "To make a Web service out of an object in a data flow, you simply open the object's property box and click the check box 'Enable as a Web service.'" To reuse an object, "you simply take something you've already created and tested, then easily reuse it in a different context."

In fact, Informatica has a new marketing message and roadmap revolving around this data integration architecture idea: Universal Data Services (UDS). Harriet Fryman, group director of product marketing, introduced UDS by first citing an IDC study that shows enterprises will be attacking an average of 25 data-intensive initiatives over the next two years. These initiatives include real-time activity monitoring, data warehousing, legacy migration, consolidation, data synchronization and replication, and reconciled single views.

The UDS architecture provides a common set of services, built on top of a shared foundational layer ("data server"). Informatica's shared data services let IT departments reuse common services for multiple tasks across all data integration efforts, including access, integration, audit, and visualization. Informatica's existing products fulfill many of these roles, but the company will introduce more services over time. The UDS data server is an engine including foundation services, such as enterprise-level performance, scalability, availability, metadata management, optimization, security, scheduling, and workflow.

SAS contends that its SAS Intelligence Platform, which supports data integration as part of a larger enterprise intelligence underpinning, is satisfying user needs without the buzzy marketing term "data services." "Our platform approach continues to expand the availability of API-level access to and from the SAS platform to speed delivery of information," says Rob Stephens, director of technology strategy for SAS.

Pervasive, which markets to midsized enterprises, began touting its integration architecture in March, soon after acquiring Data Junction. Mike Hoskins (VP and GM of Pervasive's Integration Products Division) pointed to growing integration demands and the lack of scalability inherent in traditional point-to-point integration methods. "What broke the back of the integration industry was the need to integrate customers and suppliers. We live in a world of huge variety and constant change," Hoskins adds.

What's needed is support for flexible, scalable integration, and several vendors are showing significant efforts to provide it.