May 05, 2010, 1:10 PM — Mention "canonical information model" in some circles, and people will run screaming from the room. Memories of unending quests to map the corporate information model are still fresh for these IT pros, creating a post-traumatic-stress response. Is it any wonder that formal infrastructure architecture (IA) practices have had trouble getting off the ground?
A recent Forrester survey reveals that IA domain development lags behind service-oriented architecture (SOA) adoption, with only one-third of organizations having formal IA programs. Of the shops pursuing SOA, nearly 55% are pursuing a canonical information model and 45% are going forward without a canonical information model. This means that many businesses are pursuing SOA without clear guidance from any central authority on what information sources to use for services in general and information and data services in particular.
Although it may seem daunting, the bottom line is that cross-silo business solutions such as SOA, Business Process Management ( BPM), and Infrastructure-as-a-Service (IaaS) require well-designed information architectures in order to base their activities on trusted data. Organizations pursuing SOA without well-established IA risk undermining the inherent value of the SOA business service - the availability of common services across the enterprise.
An informal approach leads to problems when a series of unconnected one-of projects - implemented across the enterprise - yield a confusing array of data services with overlapping functionality.
IA success depends on your ability to apply "street-level strategy," which combines a long-term vision with near-term investments toward that vision. Forrester first described street-level strategy as a way to approach SOA, however, the sheer size of the problem that IA targets makes it an ideal candidate for this approach. The following outlines how development teams and EA can work together to build a pragmatic IA strategy:
1. Start by creating a vision
Rather than fully documenting the current and target state with detailed models, create a vision and stop there with the initial planning documents. The vision document should include high level goals, key principles, the benefits to gain and the problems to lose, and a conceptual architecture showing the major information subject areas.
One of the better ways to present a high-level vision of IA is to map the conceptual-level IA to a high-level business capability map. This avoids the error of presenting information without a business context and sets the stage for strategic enterprisewide thinking by linking key information entities to organization-agnostic business capabilities.
2. Select and execute projects
Look for projects that focus on business access to information that can embody a well-architected approach and add to the enterprise's collective understanding of data. Rather than create yet another disconnected and redundant reporting system for a limited audience, do a limited "archeological dig" for appropriate existing data stores that you can reuse for your project. Initiate discussions with business area subject matter experts (SMEs) to fully understand the context for the information, relate that to the available information sources, and begin to capture and manage that metadata centrally.
3. Evangelize good IA practices
IA will not make progress on its own merits; it needs a champion. Forrester has found that the only effective way to market architecture initiatives is to characterize them as beneficial to stakeholders. By addressing the key concerns of the most important organizational roles, it's more likely you'll win their support. Create a compelling way to describe IA's benefits that links IA to needed business outcomes and capabilities and begin selling upward to EA management.
4. Insert early-stage governance
IA governance is more difficult than technology governance as it requires the participation of business-side roles. Prioritize your evangelizing efforts for the executives you will need to support the governance effort and the business and IT staff who need to be recruited as data stewards. It's a good idea to start with projects where the value is clear, such as those related to data warehousing, business intelligence (BI), or master data management (MDM) efforts.
5. Build regular interactions with appropriate parties
