EA Principles - October 2012
Toward a Scientific Enterprise Architecture
Published: October 1, 2012
In this third article about EA Principles, Eskil Swende describes an approach that may lead us toward a more scientific Enterprise Architecture.
1. Some important criteria to be met by a scientific Enterprise Architecture (EA):
- EA must have “traceability” toward the business solutions. That means you must be able to find out if the solution follows the constraints of the EA. Compare with the City Plan where you are able to evaluate if a new building has complied with the City Plan or not.
- EA must deliver “agility” to the enterprise, always being well prepared for changes.
- EA must be as “stable” as possible to prepare the enterprise for ongoing changes.
- Invest in “reusable” components based on the stable information resource (compare with the car industry below).
- Reduce the “dependencies” between the components in the architecture making it possible to change or add new components without affecting other components more than necessary. This will minimize the integration and increase agility.
2. Transformation to the “reusing” stage
Enterprise Architecture as Strategy is the well-known book by Weill, Ross and Robertson that really introduced EA to management all over the world when it was published in 2006. It describes the four development stages of EA from Localization (1), Standardization (2), Optimizing (3) to Reusing (4). They state that so far only 2% of our enterprises have reached the stage 4 of Reusing.
Even more important is their book IT Savvy published in 2009, which has unfortunately not attracted much attention at all. Here they explain the very important transformation to the “Reusing” stage.
Some important statements in their book relate especially to the criteria of a scientific EA.
- Data/information is the critical integrating mechanism. (That is why the Information Architecture based on Codd’s normalization rules is key in a successful EA – my comment.)
- Identifying what is stable makes way for change. (EA must invest more in stability and less in changes – my comment.)
- Data and process standards enhance innovations. (Standards give the stability that makes way for changes – my comment.)
3. The car industry has already made their transformation to reuse
Today the car industry invests heavily in modularity. VW calculates cost reduction of 5 billion Euros each year. A standard module will be reused in 40 different car models when producing 4.9 billion cars at VW. It allows many models to be produced in the same site reducing the operational cost with another 20 – 30 %!
The potential to invest in reusable components in EA is even greater. If we really want to achieve stable modules we must base them on the Information Resource. Based on Codd’s normalization theory, they will become stable, more stable than business processes or the organization. The information resource is also unique compared to most other resources because it is not consumed when used!
Comparing with the car industry, EA should invest in reusable components to be used in many business solutions. That will reduce the cost and accelerate the development speed of new business solutions. Comparing again to the car industry, you must be a business expert when developing reusable modules, and you must be an expert in information modeling to achieve a critical level of stability and quality of a reusable component. Today most enterprises are short of that capability and our universities do not focus on educating students to achieve that capability, mainly because this capability has so far not been asked for.
Today a car industry trying to develop a new car model from scratch and not reusing standard modules will not be competitive! Tomorrow an enterprise that does not invest in reusable standard components will not be competitive. They will be too slow and spend too much money in development and maintaining a complex infrastructure.
It took the car industry about 100 years to establish reusable components. Let us hope EA can make it in just 50 years!
4. Toward a useful definition of EA
“The Enterprise Architecture ensures that the enterprise is always fit for purpose of achieving its vision, mission and strategy.”
This definition is proposed by Len Fehskens in his Re-Thinking Architecture approach with my addition of the word “always.” An Enterprise Architecture is founded on a continuum, where its lifecycle starts when the enterprise is established and ends when the enterprise is finished.
Explaining the definition
When expressing the “vision, mission and strategies,” we use the definition of the “value proposition” in the Osterwalder Canvas or the definition of “customer values” in Hammer's Business Processes. They are well-defined and fulfill the criteria described below.
The small word “always” in the EA definition may be very important. That means EA must not only be fit for purpose today, but be prepared for future known or unknown changes in vision, mission and strategies. This may have a very fundamental impact on the EA as we will see below.
5. Bringing Osterwalder, Hammer and Codd together
We will illustrate this in a normalized information model describing how “why, how and what” may work together.
6. Defining the entities
Business Innovation Canvas
A Business Innovation Canvas describes the rationale of how an organization creates, delivers and captures value. One organization normally has several Business Innovation Canvases. A new Canvas may be created at any time and an existing Canvas may become obsolete, so a Canvas has a more limited lifecycle than EA has.
In a canvas there are nine basic Building Blocks showing the logic of how an organization intends to make money. The nine Building Blocks cover the four main areas of a business: customers, offer, infrastructure and financial viability. The nine Building Blocks are customer segments, value propositions, channels, customer relationships, revenue streams, key resources, key activities, key partnerships and cost structure. One Building Block may be reused in several Business Model Canvases.
A Business Process is initiated by an event and consists of a collection of activities, which together create an output of value to the customer. A Business Process does not describe the flow of activities, so it may not be mixed up with “work flow.” The Business Process consists of the activities adding value to the customers. All other activities must be omitted.
A Business Process consists of a number of activities. One activity belongs to only one Business Process.
A Business Information Group consists of a number of entities. One entity belongs to only one Information Group. An Information Group is either Master Data or Event Data. Master Data is divided into Person Data, Product Data or Infrastructure Data.
An entity is a basic concept in an organization. It may be an event, a person, a place or a thing of interest to keep information about. It has to be defined by a unique and stable identifier. A single data element or attribute belongs to one and only one entity (the normalization rule).
7. Describing the relationships
A Business Innovation Canvas relates to several Information Groups. The same Information Group may relate to many Business Model Canvases illustrating the reuse of the Information Resource. The relationships show us how Master Data may be reused in several Canvases.
In one Business Process a number of Information Groups are handled. One Information Group may be handled in many Business Processes. When establishing this relationship, it is important to avoid unnecessary dependencies and creating a complex integration structure.
In one activity, a number of entities may be handled and one entity may be handled in several activities.
8. Evaluate and select innovations based on the EA
EA Principle #7. Relate the Enterprise Architecture to the Business Innovation Models
- The business ideas described in a Business Innovation Model need to be integrated with the enterprise structure.
- The integration between innovations and EA requires a well-structured Business Innovation Model.
- Integrating EA and the Business Innovation Models proves the importance of a good enterprise structures. A way to obtain top management involvement and commitment.
- Each group of Master Data in the enterprise structure is related to one of the nine Building Blocks in the Business Innovation Model.
- Master Data is one type of Information Group in the Information Architecture. The other type of Information Group is Event Data. Event Data is normally created in the operational process and relates to the building block “key activities.”
- An enterprise may have a number of Business Innovation Models, where the same Master Data and the same Business Activity relates to several different Innovation Models.
- A Business Innovation Model is related to the Information Architecture and Business Processes.
- When the Business Innovation Models are changed, the enterprise structure needs to be changed accordingly and then transformed to the Business Solutions.
- Business Model Generation by Alexander Osterwalder, 2010
9. Continued development toward a Scientific EA
At the 4th EA Forum in Venice that took place in late August 2012, the discussions of a more scientific EA were initiated by IRM. It was inspired by Len Fehskens and his approach Re-Thinking Architecture. These discussions will continue trying to use the “why, how, what” approach describing the Enterprise Architecture itself.
If you want to take an active part or just follow the future development, just send me an email
Go to Current Issue | Go to Issue Archive
Recent articles by Eskil Swende
- Eskil is main partner at IRM
, a Scandinavian consulting company focusing on Enterprise Architecture and the
Innovation Process. He is also a partner at IRM UK
, a strategic education company in London that provides seminars and arranges yearly conferences on EA, IA, MDM and BPM. Eskil is President of DAMA Chapter Scandinavia and has developed a global wisdom network of leading experts inside and outside DAMA, inviting them to give presentations and tutorials in Scandinavia. He can be reached at Eskil.Swende@irm.se