To paraphrase the Paul Simon song, there must be 50 ways to integrate yourenterprise data. In recent years, companies have made all kinds of attempts to integrate both their applications and data - employing techniques from sophisticated enterprise application integration projects all the way down to manual hand coding. However, while most of these approaches work at least some of the time, few, if any, are delivering real agility for their businesses.
Recently, I had the opportunity to moderate aWebcast- sponsored by Informatica and hosted by ebizQ - which explored in detail an emerging approach, called data services, which ties into service oriented architecture (SOA) and creates a data abstraction layer that addresses the complexities seen acrossenterprise dataenvironments.
Leading the Webcast wereAsh Parikh,principal product marketing manager for Informatica and a highly regarded industry speaker and author, andDavid Ramos,director of business intelligence and analytics forLinkShare Corporation.
In his presentation, Ash urged closer collaboration between the enterprise data management and emerging service oriented architecture (SOA) worlds. (John Schmidt recently provided a nice overview of SOAhereat the EDM blogsite.)
Ash observed that current approaches to enterprise data management have worked well from an application point of view, but have been ineffective for enterprise data."They're the wrong tools for the right job," he said. The problem, he explained, is that while many of these solutions do a great job of addressing issues at the application layer, the data layer still remains an untamed frontier.
"There has been a proliferation of business services such as BI apps and portals," Ash explained. "There are various integration technologies out there such as EAI, which are very application centric. They do a remarkable job of handling integration challenges at the application layer." However, he continued, while spectacular progress has been made in this area, enterprise data has been a poor cousin. Current integration efforts assume, incorrectly, "that the data that has been served up to these businesses services is of the highest quality," Ash explained:
"There is little or no reuse of the data, and there are inaccuracies and inconsistencies. These services did not address various data latency requirements and data volume needs. The confusion has been sorted out at the upper layers, but still exists now at the data layer. Along with that comes poor governance of enterprise information sets. This doesn't make an effective foundation for enabling business agility."
While EAI itself is a vast improvement over manual coding of integration code, it falls short when it comes to addressing requirements within the enterprise data layer."That leaves limited extensibility or reuse across various projects," Ash said. "There's no way to know the origin of data, or how it's being used. It requires elevated levels of ongoing maintenance. There's no support for data movement, and there's no easy way to automatically detect changes in various data sources."
While SOA can pick up where EAI left off, SOA alone will not address the vexing issues of enterprise data integration. "SOA promises to deliver business agility by breaking down barriers between silos of applications, and by reusing business services," Ash said. "However, if the data stuck inside silos is bad, is stale, or is inaccurate, imagine the calamity. The silos may disappear, but then data from many different applications becomes co-mingled."
The issue isn't simply about enabling access to data across the enterprise, Ash said.The greatest challenge enterprises face is ensuring the quality of the data that becomes accessible as a result of SOA."Enterprise data is complex, it's about volume, latency, and many formats. It requires that as part of SOA, data be treated as a strategic enterprise asset that addresses the various data integration challenges."
The way to accomplish this purposing of enterprise data assets within a SOA environment is through the delivery of "data services," Ash continued. "Data services is a highly flexible simple and cost-effective solution that provides the model and standards-based reusable abstraction layers that lower the complexity of delivering data from silos. Data services deliver a single consistent view of all enterprise data at the right time."
From a technical viewpoint, a data service is a modular and reusable well-defined business relevant-service that leverages established technology standards, he adds.A data service "enables access integration to right time data throughout the enterprise and across corporate firewalls. Data services create an abstraction layer to all analytical, operational information, and serves it up to other abstraction layers, which could be an [enterprise service bus],"he says.
Many organizations today are struggling with the complexity of multiple systems and data sources. The combination of data services delivered within a service oriented architecture framework offers relief and far greater agility from the slower and more expensive methods that prevailed in the past.