2011-02-19

Explaining EA: business architecture basics 2


Note: a revised version of these three posts is available at http://www.improving-bpm-systems.com/pubs/Explaining-EA-BA-basics_v7.pdf

Continued from Explaining EA: business architecture basics 1

4 Linking WHY, WHAT and HOW


So, an enterprise’s value-chain and value-streams are the high-level decomposition of the work of the (whole) enterprise into the work of many different activities. In such a decomposition, WHY + WHAT of the whole enterprise should be used to define WHY + WHAT of each activity. The glue between them is HOW. Let’s look at a fictitious scenario.

Stakeholders:
OK, your business model looks good. Now tell us about the operating model.

Future CEO:
Our business model is the WHY for our operating model. The latter starts by showing the relationships between the enterprise and its partners (suppliers, providers, customers, etc.) from the economic ecosystem. Within the enterprise we have identified 4 aggregations of value-streams: customer-centric (green), strategic-visioning (blue), people-caring (yellow) and business enabling (red), as well as the relationships between them.

The enterprise and related external partnersThe enterprise as a set of aggregations of selected value-streams





Example from www.enterprisebusinessarchitecture.com.


We know all our value-streams and their integrated components. Each value-stream is connected to a particular objective. Also, we know our value-chain.


Value-streamsValue-chain


Example from www.enterprisebusinessarchitecture.com.


So, for each value-stream (FUNC1), we know its input WHAT0, its output WHAT1 as well as its operating requirements WHY0.

Stakeholders:
Sounds great. And, can you assure us that FUNC1 is capable of operating as required?

Architect:
The desired performance of FUNC1 is guaranteed by its implementation (HOW1) as the explicit coordination of “smaller” functions. In some way, WHAT1 is decomposed into a set of WHAT2x. WHY0 is decomposed into a set of WHY1x, and FUNC1 is decomposed into a set of FUNC2x. They are all coordinated together. In the illustration below, the coordination is trivial, but in real cases it may be rather complex (e.g. an interaction of activities carried out by several interdependent functional roles).


Stakeholders:
Please continue until all FUNC# become “manageable” activities so that they can be bought, rented, outsourced and easily implemented.

Architect:

This will involve the explicit decomposition of each value-stream to reveal the horizontal (peers) and vertical (subordinated) structure.

... Some time later ...

Architect:
As a result of this decomposition, a directed graph can be obtained (see the figure below). This directed graph is represented as a river basin; it could also be represented as an iceberg in which the value-stream is the tip of the iceberg.


In this graph, nodes (i.e. activities) are connected by edges to show the dependencies between results (i.e. the result of activity C depends on the results of activities I, K, L and B). This means that the result of a particular activity contributes to the result of another activity (which is probably more valuable and thus more expensive). The timing of result generation may be different: some results can be produced in advance and stored for later, some results can be produced on demand and some results can be acquired just before they are needed.

The primary importance of such a graph (called a “value & expenses basin” or “VEB”) is to represent business performance – the business wants to delight the customers (by giving them what they want to pay for) and the shareholders (by creating a profit). As shown in the figure below, different activities contribute differently to the generation of the value (green arrows) and the associated expenses (red arrows). The width of the arrows signifies the relative amount of value or expense.


The VEB should help in the management of an enterprise. It represents a dynamic, actual and contextual contribution of different activities to the value and expenses associated with a particular result. The business can be attentive to different “tributaries” which are

a) the most value-adding,
b) the most wasteful,
c) doing worse than defined by WHY, and
d) doing better than defined by WHY.

Depending on the business needs, such a representation can display a particular instance of value creation or a set of instances (usually over a given period of time).

So, how is a VEB constructed?

A VEB is not a flow of control, an event processing network (EPN) or a PERT diagram. It can be considered as a flow of assets (or a data flow diagram), but this will be just an externally-visible representation of internal mechanisms. Such a representation is good enough for the reactive analysis of behaviour, but is not sufficient for active control and pro-active (predictive) analytics. It is necessary to have a dynamic model which can be used for execution (e.g. simulation) and from which the VEB can be generated.

The set of “internal mechanisms” (as mentioned above) is a superposition of different coordination techniques (token-based, rule-based, event-based, data-based, etc.) as illustrated in the following.
  1. An activity from one value-stream (or business process) can obtain some assets (business objects) which belong to another value-stream (or business process). This is pull-like communication, e.g. the “Order-to-Cash” value-stream should know the customer’s address which is maintained by the “Prospect-to-Customer” value-stream.
  2. An activity from one value-stream (or business process) can send some assets to another value-stream (or business process). The latter interprets appearing of the assets as an event to be treated. This is push-like communication. Usually, there are three ways in which this treatment can occur:
    • a new instance should be started (e.g. for the manufacture of something) – initiating event;
    • an existing instance, which is waiting for this event, consumes the event and continues its work (e.g. the confirmation of a payment) – solicited event;
    • an existing instance, which does not expect this event, has to react to it – unsolicited event.

In reality, the situation is rather complicated. An enterprise may have several value-streams running in parallel. Some activities can be shared between different value-streams and some value-streams may compete for limited resources. Some activities may be outsourced or insourced, etc. All of these complexities need to be taken into account.

Furthermore, in addition to the activities, there are several other artefacts (see chapter 6) which should be defined explicitly in the model.

Continue at Explaining EA: business architecture basics 3

No comments: