|Bottom-Up Approach (Point Projects)||Top-Down Approach (Area Projects)|
|Local short-term initiative. No mandate to pay attention to broader, longer-term opportunities and effects.||Broader, longer-term initiative|
|Building a solution against immediate requirements (where “building” means design, construct or assemble)||Focus on system properties across a whole area (e.g. business domain, technical domain, infrastructure)|
|Strongly aligned to local objectives. Direct link between (local) benefits, costs and risks.||Indirect links between benefits (across area), costs and risks
|Cost-effective use of conveniently available resources (improvisation or “bricolage”)||Creating value by establishing (procuring or building) conveniently available resources|
One way of addressing this dilemma is to introduce a twin track process, involving a top-down stream of activity and a bottom-up stream of activity.
Obviously for this twin-track process to be effective, we need clear allocation of responsibility, authority, expertise and work (RAEW). This is an aspect of governance - making sure the right things are done in the right way. Twin-track development exposes the inevitable tensions between business goals and service needs. And in federated/distributed development, these tensions are replicated across multiple business entities; governance then becomes a question of negotiation between two separate organizations, rather than simple management resolution within a single right framework.
This is a modified extract from an article on Business-Driven SOA published in the CBDI Journal, June 2004. For further extracts from this article, please see my Slideshare presentation on Organic Planning. See also our papers in the Microsoft Architecture Journal on Metropolis and SOA Governance (July 2005) and Taking Governance to the Edge (August 2006).
The notion of twin-track development is included in the Practical Guide to Federal SOA, published by the CIO Council in 2008. See also
- Stuart Charlton, On Effective Architecture (October 2007)
- Jeevak Kasarkod, CBDI SAE Reference Framework Continues to Evolve (InfoQ November 2010)
- Dave Mayo, Agile IT (June 2011)
- Mike Papazoglou, Engineering Services (??)
- Ronald Schmelzer, Why Service Consumers and Service Providers Should Never Directly Communicate (Zapthink, Feb 2008) with my response on Agility and Variation (March 2008)
- David Sprott, Beware the new silos (March 2011)
For other posts on twin-track: browse, subscribe.