Long-Tail SOA and the Mythology of Re-Use
In praise of the niche market, the other 80%
By: Marc Rix
May. 22, 2008 10:30 AM
As said, different business customers need different IT services. When a service is needed by more than one customer reuse happens. Reuse is a measure of the demand for an individual service and is generally good. But it’s also important to measure the demand for entire collections of services, or service “markets.” The aggregate demand for mainstream services is huge, encompassing roughly 80% of all service consumers. By contrast, the aggregate demand for niche services is relatively small at 20%. At first blush this might suggest that building niche services makes poor business sense. But due to the Long Tail effect, four niche services are in demand for every one mainstream service. In other words, mainstream SOA leaves four markets underserved for every one it satisfies. These four markets may be small, but they still represent a missed opportunity for IT. How much missed opportunity is a function of the protection those services provide during times of change.
The value of SOA is revealed during turbulent times. This is where the SOA rubber meets the road. Like the distribution of service demand, the distribution of change across the enterprise isn’t uniform. Some areas of business are more susceptible to change than others, and some periods of change are more disruptive than others. SOA strategies should be in tune with these differences, otherwise they risk committing the cardinal sin of IT – being unprepared to facilitate fluctuating business needs.
IT must be prepared to deal with the magnitude of change. We have a good understanding of this already by virtue of our mainstream IT mentality. We know first hand that when change affects core enterprise systems – through government regulation, acquisitions, legacy system retirement, and so on – the shockwave can rip through the entire company. Being unprepared for major infrastructural change leaves a big crater and can be the death knell for a company. Mainstream SOA protects against this kind of threat. By contrast, a change to a niche service leaves a small crater and only affects a small pocket of the business. Companies usually heal quickly from isolated niche threats.
IT must also be prepared to deal with the frequency of change, which is often overlooked. Every time change, big or small, is allowed to affect the business, the business’s defenses are weakened. Mainstream SOA guards against the biggest threats but lets the smaller ones through, which would be fine if there weren’t so many small threats to worry about. In reality, though, big change is rare and small change is constant. Mainstream services change infrequently because they are mainstream services. They are owned by IT, they are the lifeblood of the business, and carry core functions and data to all facets of the enterprise. They uphold the fundamental policies, business practices, and values of the company throughout the business ecosystem. They must remain as stable as possible because changing them risks disrupting the business on a global scale. As such, change in mainstream IT typically occurs on time scales measured in fiscal years or quarters.
Change in niche IT, on the other hand, is measured in days. Niche services are owned by the business and change with external market conditions. They deliver data and functions that provide relevance and differentiation to business units in local markets. People at the edge of business – sales reps, regional managers, product managers, R&D engineers – need more than mainstream one-size-fits-all services because their customers need more than generic one-size-fits-all products. These areas of the business make demands of IT that are a direct reflection of their customers’ needs, which are always changing. For IT not to respond to small changes in niche service markets is to allow the systematic erosion of LOB revenues. Real IT-business alignment and agility comes from Long Tail SOA because when change happens, it’s most likely to be felt at the front lines rather than the back office.
Which is the best SOA strategy? In the end, is it better to hedge against infrequent big change or frequent small change? Is it better to provide services to a few large audiences or many small ones? Is it better to strive for reuse or agility? Like most situations, posturing at one extreme or the other is seldom optimal. The best strategy is the one that best addresses the strategic goals of the company, but probably involves some amount of blending of mainstream SOA and Long Tail SOA.
Make Long Tail SOA an extension of mainstream SOA. Most niche services are refinements or customizations of mainstream services. A mainstream “Get Property For Sale” service could evolve into “Get Houses By Zip Code In Price Range” as it travels down the Long Tail. In this example, the original service became less valuable (generic and reusable) at the mainstream level but more valuable (customer-oriented) at the niche level. Although the second service is a different “product” than the first with a different market, it doesn’t have to be implemented separately. In fact, it should borrow as much as it can from the generic service and just add the unique parts. For example, it could call the generic service and refine the results based on its specialty. In other words, Long Tail services can be derived from mainstream services simply through effective service composition. Easily stated, but this requires a bit of discipline.
Hybrid SOA requires a strong Mainstream SOA foundation. Because Long Tail services are dependent on mainstream services, mainstream services need to be stable. If a mainstream service changes it could cause entire families of related Long Tail services to change. Solid design, high availability, and loose coupling are essential at the foundation service level.
All services must be discoverable. In Hybrid SOA, single-service solutions are rare. Instead, IT solutions are value chains of services of varying granularity and complexity strung together to meet specific business needs. Developers (or mashers!) of Long Tail services need to know where the raw materials are to draw from. Otherwise, the services either can’t be built or are built by duplicating existing services.
Finally, strong standards and governance are required to enable hybrid SOA. This point may seem obvious, but the importance of standards and governance should be stressed since hybrid SOA brings two distinctly different service economies together. Standards and governance practices that are effective in either mainstream SOA or Long Tail SOA may not be relevant when the two are combined. For instance, mainstream SOA may operate well without strict interoperability standards, especially if the major IT systems are more or less homogenous, such as in predominantly Microsoft or Oracle shops. But interoperability standards would become important when introducing Long Tail SOA if technology used by lines of business were more of a mixed bag. IT may also not have the resources or bandwidth to supply all Long Tail services, in which case niche service development (or mashing!) would need to be decentralized. Standards and governance would need to be rigid enough to maintain quality of output yet flexible enough to accommodate a diversity of composition tools and methodologies.
The Bottom Line
Neither mainstream SOA nor Long Tail SOA is effective alone. They are complementary strategies that should be blended together to maximize the benefits of Service Oriented Architecture. In hybrid SOA, mainstream services provide the foundation for enterprise-wide networks of service-oriented value chains. These value chains flow down into the Long Tail and deliver specialized just-in-time services to small teams at the edges of business – where agility matters most.
Reader Feedback: Page 1 of 1
SOA World Latest Stories
Subscribe to the World's Most Powerful Newsletters
Subscribe to Our Rss Feeds & Get Your SYS-CON News Live!
SYS-CON Featured Whitepapers
Most Read This Week