Guest View: Business Intelligence in the Age of SOA

By Charles Nicholls

December 15, 2007 —

History shows that wherever there are software applications, business intelligence follows. It used to be possible to link BI to a database or data warehouse in order to analyze company performance. Companies would, and still do, try to shuttle increasing volumes of information into data warehouses, and then extract it for analysis.

Getting data into and out of the data warehouse turned into a complicated chore in its own right, but in the past few years, it’s become overwhelming as applications have proliferated and become more sophisticated. Most businesses recognize that they need to analyze this information if the best decisions are to be made, but are still applying retrospective BI technologies and approaches to the problem.

The most elemental challenge to traditional BI is the requirement to analyze data as part of a business process, not simply to report on it after the fact. In order to build BI into processes, BI needs to be real-time. This represents a seismic shift for an industry that runs on batch updates.

The old architectural approach to BI—adding it after applications have been built and focusing it on the database or data warehouse—doesn’t make sense in an event-driven world. At the same time, those events, in a service-oriented architecture (SOA), represent an unprecedented opportunity for analysis and action.

Faster Isn’t Fast Enough
The new approach to BI is enabled by the flow of data through applications and middleware. This means that companies building a SOA can benefit from determining now how BI can help them get the most out of their investment. Indeed, companies have to add BI to SOA—it simply becomes a question of how and when they’ll do it.

The traditional route to BI is database-centric and focuses on giving companies dashboards for looking at information once it’s made its way into the data warehouse. But data warehouses are out of date. Extracting, transforming and analyzing information that’s even hours old doesn’t tell the company what to do right now.

It’s not that the data warehouse no longer matters in a SOA; it does. But it should be understood for what it is—the system that affords a look at past performance. So if companies are looking to the data warehouse to provide actionable information, they’ve made an elemental mistake.

To gain insight, and to achieve the benefits SOA promises, businesses need to view BI differently. There are a few approaches that generally get considered.

The first is BI as a Web service. Some analysts have described the traditional BI vendors as “sleepwalking into SOA,” and the vendors have reacted by providing Web services interfaces. The problem is that the data is usually coming from a data warehouse and therefore is old. It also doesn’t contain process state data, so using BI as a service is useful only for historical data lookup tasks. Companies can wire BI tools to operational systems, but this introduces a performance impact.

Next is data as a service. Several middleware vendors position their distributed query platforms as “data as a service.” Here the concept is to offer one interface that will provide a heterogeneous join on data retrieved from databases. Most of these systems have a cache built in so that repetitive data lookups are faster. This can work well for slowly changing dimensions, but as soon as significant volumes of data are involved, performance suffers.

Finally, there’s event intelligence. This approach doesn’t rely on queries and therefore has no impact on operational data sources. Rather, an event intelligence approach uses events flowing through the SOA infrastructure or published by BPM tools as its data source. Architecturally different from traditional, query-based BI, event intelligence maintains continuous calculations in real time, enabling complex calculations to be built into business processes.

Considering these approaches is useful when weighing SOA implementations. The more fundamental question is: When should companies consider the different approaches? In most SOA approaches, BI is still an afterthought, but this is changing rapidly.

Intelligence From the Ground Up
Industry research firm Ovum suggests that approaches to SOA today fall into three categories:

First is the clean slate, in which an organization with no legacy code can design a SOA from scratch. Also in this category are organizations that take a strategic approach and evaluate their demands for information before they consider the technology involved.

The second approach happens when companies view SOA as a development pattern and implement it in a technical manner. Almost invariably, the business doesn’t understand what SOA is and conversations about the types of intelligence users across the business need to receive haven’t happened.

There’s also a middle-ground approach to SOA, in which a line-of-business sponsor works with IT to solve a defined problem and includes BI to analyze certain anticipated questions. SOA at the departmental level can be useful in some cases, although it doesn’t address dynamics that push people to work cross-functionally.

According to Aberdeen Research, the biggest challenge in corporate IT is imparting real-time visibility into business operations. Indeed, gaining that insight is one of the major justifications used for most SOA projects. Software professionals must recognize, though, that simply building a SOA will not make processes more intelligent. Unless event intelligence is built in, the new architecture will simply automate dumb processes.

Focus on Middleware
The most elegant approach to adding event-driven BI to a SOA environment is to integrate it from the outset. The trend toward middleware-oriented development paves the way by promoting the construction of loosely coupled services that deliver flexible applications. Now, rather than reporting on processes after they occur, BI integrated can support decisions in real time. The new BI applications do this by working with the services that help create the SOA.

Once the SOA has been built, however, bolting BI on after the fact in a loosely coupled environment becomes an architectural mess. In effect, IT must reintegrate services that had been loosely coupled.

Event Intelligence for Smarter Decisions
The answer is to build in event intelligence from the outset of the SOA project. In this way, the business keeps the flexibility of the SOA and allows applications to be altered or added as business needs change—all while gaining access to the data needed to power smarter decisions.

The ability to make real-time decisions opens up new ways of doing business and interacting with customers. The first step to getting there is to talk with business users and ask what they want to measure, and then to use the new generation of BI tools to deliver that information.

Does this mean throw out the data warehouse? Certainly not—it remains the system of record. But it does mean that with SOA, you’re going to be able to use BI in an entirely new way.

Charles Nicholls is founder and CEO of SeeWhy Software, which sells business intelligence software.

Share this link: http://www.sdtimes.com/link/31439

SOA Software releases project-planning suite for SOA transition Portfolio Manager provides a framework for SOA planning, helping developers prioritize services, understand dependencies, and plan architecture and governance processes, the company says. The product is marketed as being essential for creating road maps for transitioning to SOA.

Red Hat prepares complex events processing for SOA Updates to JBoss Enterprise SOA Platform and JBoss Operations Network lay the groundwork for enhanced CEP and rules management. Beyond that, new security features have been implemented in Enterprise SOA Platform.

SOA’s dead; long live SOA While software tool providers can no longer pin their products to the term, its best practices live on inside the cloud

Add comment

Name*
Email*
Country

Comment
Preview

News on Monday
more>>
SharePoint Tech Report
more>>

Download Current Issue
ISSUE 3/15/2010 PDF

Need Back Issues?
DOWNLOAD HERE

Receive the print Edition?

Google Code turns 5
Google Code Turns 5, and adds a Paxos Algorithm to make the system more stable and reliable.
03/17/2010 11:16 AM EST

Test your Visual Studio 2010 know-how
Microsoft is offering free beta certification exams for Visual Studio 2010.
03/17/2010 11:08 AM EST

Microsoft lifts the hood on IE9
Microsoft is previewing IE9.
03/16/2010 01:10 PM EST