Is Your SOA Hammer Looking for a Nail?
It sounds so obvious when you get right down to it: you need to know what problem you’re solving before you can solve it. Common sense tells you to start with the problem before you can find the solution to the problem. If you start with a solution without knowing what the problem is, then there’s no telling if the solution you have will be the right one for the problems you’re facing.
Obvious, yes, but it never ceases to amaze us at ZapThink that when it comes to Service-Oriented Architecture (SOA) projects, time and again we run into technology teams who don’t have a grasp as to what business problems they’re looking to solve. Now, it might be tempting to dismiss this disconnect to “techies being techies” or some other superficial explanation, but the problem is so pervasive and so common that there must be more to it. As a result, we took a closer look at why so many SOA projects have unclear business drivers. What we found is that the underlying issue has little to do with SOA, and everything to do with the way large businesses are run.
The Wrong Question
This story begins with the SOA architect. Architects frequently call in ZapThink when they’re stuck on some part of their SOA initiative; we’re SOA fire fighters, so to speak. Frequently, the first question we get when we sit down with the architecture team is “how do I sell SOA to the business?” Well, if that’s the first question they ask, they’re already on the wrong foot. That’s the wrong question! The correct question is “how do we best solve the business problems we’re facing?” SOA is not a panacea, after all; it only helps organizations solve certain problems typically having to do with business change in the face of IT heterogeneity. It’s best to solve other problems with other approaches as appropriate, a classic example of the right tool for the job.
For this reason, ZapThink considers the SOA business case as an essential SOA artifact. Architects must have a clear picture of the business motivations for SOA, not only at the beginning of the initiative, but also as the architecture rolls out. After all, business issues evolve over time, partly because of the changing nature of business, but also because properly iterative SOA efforts solve problems during the course of the overall initiative.
Even when architects ask the right question, however, there is still frequently a disconnect between the business problems and the SOA approach. The challenge here is that the architects — or more broadly, the entire SOA team — are only one part of the bigger picture, especially in large organizations. In the enterprise context, how the business asks for IT capabilities in the broad sense is often at the root of the issue.
SOA Business Driver Pitfalls
Here are some real-world examples of how we’ve seen the issue of unclear business drivers for SOA play out in various enterprises. We’ve generalized a few of the details to protect the innocent (and the guilty!).
The SOA Mandate at the US Department of Defense (DoD)
- — In aggregate, the DoD is perhaps the largest implementer of SOA in the world, in large part because they have an organization-wide SOA mandate. It’s true they have high-level drivers for this mandate, including increased business agility and a more cost-effective approach to siloed IT resources. But those general drivers don’t help much when a defense contractor must create a specific implementation.
This particular DoD project involved a contractor who delivered a perfectly functional SOA implementation to their client. The client, however, found it to be entirely unsatisfactory, and the entire multi-million dollar effort was canceled, the money down the tubes. What happened? The problem was a disconnect between high level business drivers like business agility and specific business problems the agency in question wanted to solve. The fact the client wanted to “do SOA,” so the contractor “delivered SOA” to the client only made the situation worse.
The most important take-away from this fiasco is that SOA wasn’t the problem. From all the information we gathered, the contractor properly designed and implemented the architecture, that is, followed the technical best practices that constitute the practice of SOA. In essence they built a perfectly functioning tool for the wrong job. Fundamentally, the client should have specified the problems they were looking to solve, instead of specifying SOA as their requirement.
SOA for the Enterprise Morass — ZapThink recently received a SOA requirements document from a financial services client who asked us to review it and provide our recommendations. This document contained several pages of technical requirements, but when we got to the business requirements section, it was simply marked “to be determined.” Naturally, we pointed out that among the risks this project faced was the lack of a clear business case.
Our recommendation was to flesh out the business case by discussing business drivers for the SOA initiative with business stakeholders. In response, the client told us that it wasn’t feasible to speak with the business side. The entire SOA initiative focused on the IT organization, and the business wasn’t directly involved.
We pressed them on this point, explaining how critical having clear business drivers is for the success of the SOA initiative. Why is communicating with the business such an issue for them anyway? Were they afraid of such interactions? Did they not know who the stakeholders were? Or perhaps there was no business motivation for the project at all?
The problem, as it turned out, was more subtle. They described the challenge they faced as the “enterprise morass.” As happens in so many very large organizations, there are no clear communication patterns connecting business and IT. Yes, there are business stakeholders, and yes, business requirements drive IT projects in a broad sense, but there are so many players on both sides of the business/IT equation that associating individual business sponsors with specific IT projects is a complex, politically charged challenge. As a result, the SOA team can only look to the management hierarchy within IT for direction, under the assumption that at some executive level, IT speaks to the business in order to get specific drivers for various IT initiatives, including the SOA effort. Speaking directly to business sponsors, however, is off limits.
The SOA Capability Conundrum — The chief SOA architect at this European firm was in one of our recent Licensed ZapThink Architect Bootcamps, and when we got to the Business Case exercise, he explained that in their organization, the motivation for SOA was to build out the SOA capability. His argument was as follows: if we implement SOA, deploying a range of Services that can meet a variety of business needs in the context of a flexible infrastructure that supports loose coupling and reusability, then we’ll be well-positioned for any requirements the business might throw at us in the future.
The reasoning behind this argument makes sense, at least on a superficial level. Let’s build several Business Services that represent a broad range of high-value IT capabilities in such a way that when the business comes to us with requirements, we’re bound to be able to meet those requirements with the Service capabilities we’ve deployed ahead of time. The business is bound to be ecstatic that we planned ahead like this, anticipating their needs before they gave us specific requirements!
While this organization might very well get lucky and find that they built cost-effective capabilities that the business will need, taking this “fire-ready-aim” approach to SOA dramatically increases the risks of the initiative. After all, it’s difficult enough to build reusable Services when you have clear initial requirements for those Services. Building such Services in the absence of any specific requirements is just asking for trouble.
The ZapThink Take
If you take a close look at the three scenarios above, you’ll notice that the stories don’t really have to be about SOA at all. You could take SOA out of the equation and replace it with any other IT effort aimed at tackling enterprise-level issues and you might still have the same pitfalls. Master Data Management, Customer Relationship Management, or Business Process Management are all examples of cross-organizational initiatives that might succumb to the same sorts of disconnects between business and IT.
At the root of all of these issues is the dreaded phrase “business/IT alignment.” It seems that the larger the organization, the more difficult it is to align IT capabilities with business drivers. Sometimes the problem is that the business asks for a particular solution without understanding the problem (like the DoD’s SOA mandate), or perhaps a combination of politics and communication issues interfere with business/IT alignment (the enterprise morass), or in other cases IT jumps the gun and delivers what it thinks the business will want (the capability conundrum). In none of these instances is the problem specific to SOA.
SOA, however, can potentially be part of the solution. As ZapThink has written about before, the larger trend of which SOA is a part is a movement toward formalizing the relationship between business and IT with flexible abstractions, including Business Services, Cloud capabilities, and more. If you confuse this broader trend with some combination of technologies, however, you’re falling for the straw man that gave rise to the SOA is dead meme. On the other hand, if you take an architectural approach to aligning business drivers with IT capabilities that moves toward this new world of formalized abstraction, then you are taking your first steps on the long road to true, enterprise-level business/IT alignment.