What is holding back effective data management as part of service oriented architecture?  As SOA enters a new phase of maturity, too many companies may still be overlooking its “architecture” aspect (the “A” in SOA), and therefore not realizing all the benefits that can be attained.

There is still a tendency to view SOA as a technology initiative, and thereby miss its most crucial role in the business – making information accurate, timely and accessible. My Perspectives community colleague Ash Parikh, who is also Informatica’s tireless advocate of quality data services, observes that architecture – “the most foundational piece of SOA” – is often absent from SOA projects as companies attempt to pile on solution after solution.

Ash was part of a recent Webcast panel discussion that explored the latest opportunities and challenges as SOA enters the cloud era.  The panel included Randy Heffner of Forrester; Larry Mieldezis of Liaison; and Bob Kemper of PivotLink, joined by Information Management’s Eric Kavanagh and Jim Ericson.

There are still too many instances of swivel-chair architecture in businesses, Ash said. He recounted how he observed customer service representatives on calls stalling for time by making small talk while they attempt to pull data on his account from different sources. “I know exactly what’s going on in the background,” he quipped. They are trying to figure out which ‘Ash’ they’re talking about, and basically trying to marry my information across these sources. What they bring back potentially is information — but is it consistent?  Do they have the freshest data they probably should be looking at? Is that data without any inconsistencies or inaccuracies?  Typically not.  So we still have a huge problem as far as SOA is concerned.”

Panelists agreed that the most critical issue that many SOA-aware enterprises are now running into is the viability of data being made accessible across the enterprise. Randy Heffner observed, for example, that Forrester’s surveys were finding that two-thirds of all companies implementing SOA-based infrastructures were doing so “for pure data access.”

An impressive number of companies have not lost sight of the longer-term ultimate objective of SOA, which is business transformation, he continued.  “There’s a growing realization that the business design perspective on SOA can give you transformation and agility. Whereas if they’re looking at it as just a way to push data around between applications, they don’t have the view and they’re stuck in the technology.”

However, many companies have tended to dive into SOA efforts technology first, ending up with less-than-satisfactory outcomes. For example, Randy recounted the example of one large telecom that pursued a Web services versus a more comprehensive SOA strategy, “and that got them from a place to where they had a mess of disconnected interfaces to where they had a mess of disconnected standards-based interfaces.”

Ash provided an example of a large financial services company that decided to pursue SOA: “They went around attending every industry event, picked up datasheets on every single vendor, bought an ESB, bought a registry, and bought every single thing that has SOA in the data sheet.” But five years later, still no agility and flexibility – all they had was a pile of technology.

It was time for the company’s SOA proponents to take a step back and think about what they really wanted and needed from the effort, Ash related. They needed to look beyond simple “modernization and agility” – and decide what the business needed. “Their end goal was improving the output of their customer service representatives. At that point in time, their representatives were basically working on getting a masters degree in how effective they were in talking to customers while they sat though various screens attached to various information sources, trying to get the single trusted view of their customer information.” The key is developing an architecture that supports the effective sharing of data, he said.

While SOA is an effective method for enabling cross-enterprise and inter-enterprise access to data and applications, data quality is a major issue that is cropping up, Ash observed. “What if the data sources had dirty data or consistent data to start with?” he asked. “Then you’re propagating that dirty data, up to the end point which end-users will consume that information.”

Ash advocates the creation of a data services layer as part of a service oriented architecture to manage the often-complex challenges that accompany managing data from disparate parts of the enterprise. Part of this complexity comes from the need to maintain and leverage two types of data – historical and operational. Effective decision making requires integration between both analytical and real-time information, he added. “You need a data services layer that can complement SOA. You need good data that you could trust, at a time you need it.”