As the real-time enterprise adds tactical decision making to its business intelligence (BI) toolbox, building a standardized enterprise infrastructure that can deliver transactional, tactical and strategic information in a reliable, integrated and timely way is critical. Over the past two years, according to the report “Trends in Performance Management Spending, 2006” by AMR Research, while spending on analytics infrastructure to support corporate performance management hovers between 19% and 20% of the total performance management dollars spent, a full 75% of companies surveyed are involved in some level of performance management standardization across their business.

“While the infrastructure spend has declined slightly in 2006 from the year before, according to our research—by 1% of the total percentage—this figure is belied by the increase in the number of questions we’re getting lately from companies as they plan to make data infrastructure a core part of their long-term BI architecture going forward,” says John Hagerty, vice president and research fellow at AMR Research. “Based on the number of questions we’re getting, issues surrounding the normalization of data and the standardization of infrastructure for analytics, business intelligence and performance management are at an all-time high.”

That said, if the number of questions is high, so too are the implicit challenges that companies face, both organizational and technical. “People are actually a bit more familiar with some of the technical challenges to building and standardizing their enterprise infrastructure, but those are not the hardest ones they have to face,” says Sid Adelman, a principal of Sid Adelman & Associates. “Addressing the human side of this is much more difficult. The human challenges include developing standard overall strategies for information management, doing a proper return on investment [ROI] assessment to justify the cost, providing and maintaining upper-management support, addressing the need for data stewardship and the politics of data ownership, establishing open and ongoing communication between business and IT, and providing metrics to evaluate the real business value, among others.”

Human challenges
At the core of the human challenges is the need to manage information cross-functionally. Gartner terms this need enterprise information management (EIM). According to Gartner, EIM is a formulized program for structuring, describing and governing information assets—regardless of organizational and technological boundaries—to improve operational efficiency, promote transparency and enable business insight.

Additionally, EIM is the way an organization “operationalizes” an information infrastructure to meet the needs of a real-time, agile enterprise. Enterprise information management is a holistic approach to managing information as an asset, but it also provides the roadmap for the longer-term strategic goal of converging all content (structured, semi-structured and unstructured) to meet the broader information demands in service-oriented architectures (SOAs).

“Gartner has identified seven essential building blocks to managing information as a strategic resource across the enterprise,” says David Newman, vice president at Gartner who leads the firm’s research on EIM. “These include vision, strategy, governance, organization, processes, information infrastructure and metrics. Taken together, these provide the framework organizations require as defined in their enterprise architecture strategy, as well as the information blueprint an organization will require when moving from tightly coupled to loosely coupled systems.” Without a focus on EIM, Newman adds, organizations moving to an SOA will have to deal with the proverbial Pandora’s box of embedded business rules, conflicting formats, poor data quality, incomplete metadata and other data issues that otherwise impede present-day integration efforts.

“All the building blocks have practical consequences. For example, if you want agility in your business and that’s your vision, then building in improved sense-and-response capability will be important, so you’ll need real-time information flows, master data management (MDM), semantic reconciliation and closed-loop analytics,” Newman says. “Similarly, a strategic commitment to EIM requires senior leadership, projects and budgets. Key to all this is to develop the proper governance processes and organizational structures, which complement one another.”

Governance addresses the development, maintenance, communication and enforcement of EIM policies, as well as procedures and decision rights across the enterprise, while the proper organizational model implements the governance processes. Practically speaking, this means the EIM organization needs personnel for such things as enterprise and project-level modeling, metadata and MDM, and infrastructure and data-quality management.

One process that should be firmly in place, according to Newman, is data stewardship, which addresses compliance requirements for the increased accountability and transparency of information that’s needed across the organization. Furthermore, metrics must be established to evaluate data quality, redundancy and reuse practice, as well as to measure the level of standards adoption across the application development lifecycle. “Metrics are critical,” Newman says, “and users have to be kept in the development loop to ensure their requirements are met and their accountabilities can be formalized in order to deliver business value and transparency across the organization.”

Technical challenges
When determining the technological components for your EIM program, consider a reference model or architecture to build your information infrastructure. The information infrastructure includes your data warehouse and analytics applications. The information infrastructure must illustrate the kinds of technology that will be used to integrate data, application services and business processes. Key to adopting the architecture—and to discerning the technical challenges—is to understand the relationship between transactional and analytic or decision-making services, and how both are delivered to enterprise users in a real-time services-oriented framework.

“When you start to talk about putting analytical data in front of operational frontline users to be able to support their needs, then all of a sudden the whole game changes,” Hagerty says. “New and diverse technical challenges arise for business and IT leaders who are building a real-time enterprise information infrastructure.”

Some of the technical challenges Hagerty identifies include:

Data availability, reliability, granularity and freshness. How available is the data that will be pulled from the transactional systems like SAP? What are the trusted sources? How granular or summarized are the data feeds to the active data warehouse? How fresh does the data need to be from the active data warehouse to meet operational business requirements?

Extract, transform and load (ETL) requirements. Are enterprise-wide normalization, standardization and semantic/MDM practices in place as core enterprise data warehousing competencies?

Performance. How quickly is data retrievable from the data warehouse when performing tactical queries alongside strategic analytic queries? Have you performance-tested for the increased data volume and the many more operational users who will be using the systems? What are your benchmarks?

The user loop. How flexible and responsive is the data warehouse team to new user requirements, solving real business problems and providing real value along the way, while managing change, growth and “scope creep”?

Adherence to open standards. How prepared are the development teams to adhere to industry open standards to maximize deliverability, reuse of elements and interoperability?

Evolution to SOA. Is the infrastructure being built in a layered, services-oriented way and therefore enabling development of “reusable” application components for increased access to information and decreased cost in development?

“The last two requirements are critical in that organizations really need to be able to integrate components within the infrastructure, in addition to integration of the data sources and their delivery to the end user,” Hagerty notes. “That’s the core challenge in a service-oriented architecture. For example, the fact that data has been aggregated into an active enterprise data warehouse (EDW) doesn’t mean it has a single purpose; it means it can be used in multiple ways, as a service.”

Prepare for the future
Indeed, the whole idea behind an SOA is that it provides common services and reusable componentry, so that regardless of whether that data is used to support a transaction process, a tactical decision, an analytic need or a combination of the three, it’s based on an architecture that allows for maximum flexibility, reusability and interoperability. “Though SOA is a three- to five-year vision for most people, it’s where it’s all going,” Hagerty says. “Because that’s true, building a standardized infrastructure to support the EIM vision has never been more important.” T

Michael Miley is a freelance writer in Sonoma, CA.

Why infrastructure standardization matters As businesses aggregate data across the enterprise, including within the data warehouse, dirty data is more exposed to scrutiny. There’s a tremendous drive to clean up the information infrastructure in order to really analyze operational performance with reliable data. It’s one of the reasons master data management (MDM) is such a key topic today. But the effort to standardize the data is not just about reference data or even about data warehousing, per se, but really about standardizing the data across the business. Though enterprise data warehouses (EDWs) can act as early warning systems—they help you realize that your data isn’t as clean as you thought it was—the need for an enterprise-wide plan and a standardized infrastructure arises because you’re pushing that data from the EDW to the frontlines and across the whole enterprise where everyone can see it. “What we see when people start charting this out is that they go through a fairly standard maturing process around building a standard data infrastructure,” says John Hagerty, vice president and research fellow at AMR Research. “First, they try to solve a single problem in isolation. Next, they tend to solve multiple problems in isolation. Finally, they start to realize if they continue with this one-off project approach going forward, they’re basically going to have contradictory data proliferating throughout the business. That’s when they say, ‘Stop the insanity!’ and begin to recalibrate their whole plan for how they need to use this data across the business. They realize that standardizing the data infrastructure—often using the data warehouse as the fulcrum while adhering to an enterprise-wide service-oriented architecture [SOA] as the map for the project going forward—is the next big thing that they need to do.” It’s the next big thing because, once you’ve generated a consistent source of standardized and cleansed data in an EDW, it can become the wellspring for everything you do, Hagerty explains. Anything you want to do in your environment that needs access to really standardized and cleansed data can use this common framework. You have consistency of data and consistency of understanding, because all of it is now governed by the same set of rules. That’s where SOA comes into play. “SOA means that the aggregated data in the warehouse can now be used in multiple ways,” Hagerty says. ”What you’re doing is taking it to the next step, where data provisioning is turned into service, which allows for its maximum reuse and interoperability.” Indeed, regardless of what that data is, under SOA it’s all based on a consistent source and an integrated architecture—one that allows for openness, reusability, interoperability, standardization and flexibility. “That’s the SOA vision,” Hagerty says. “Though it’s still in the future for most people, it’s critical to any forward-looking infrastructure standardization project being done today.” —M.M.

Teradata Magazine-December 2006