Just as software developers build better applications through modeling, business analysts improve business processes by creating models of process flow. In this article we'll look at how IBM WebSphere Business Modeler helps turn business processes into models that can be analyzed, simulated, refactored, and ultimately turned into software applications.

Business Modeling 101
Business process modeling documents the tasks and workflows of a business process. The first step is to determine the activities, business items, and resources required to execute the process. WebSphere Business Modeler requires little technical skill, so domain knowledge can easily be turned into a detailed process model.

A process model looks like a flowchart. It describes the sequence of steps in a business process (whether they are performed by people or automated systems). Models are visually described as a sequence of tasks and decisions linked by connectors (see Figure 1). A model can contain multiple branching paths based on decisions made during process execution. It also describes the resources available for use in the model and the cost of using those resources.

Figure 1. (Click for larger image.)

WebSphere Business Modeler also provides a variety of analysis functions that extract targeted information from one or more elements within the model (static analysis) or from a simulation of the model (dynamic analysis).

Process Modeling
To make the most of WebSphere Business Modeler it's important to understand the most commonly used process model elements. These are tasks, decisions, and classifiers.

Tasks
Tasks are the basic building blocks representing activities in a process model. Each task performs some function. Visually, a task represents the lowest level of work you can portray in a process (see Figure 2). Tasks are atomic actions, in contrast to processes, which may be decomposed into another flow. Cost and revenue estimates can be associated with tasks so that dynamic simulation of the process generates income and expense projections.

Figure 2. (Click for larger image.)

Decisions
A decision routes inputs to one of several alternative outgoing paths. You can think of a decision as a question that determines the exact set of activities during the execution of a process. Questions might include: "What type of order?", "How will the order be shipped?", and "How will the customer pay?"

There are two types of decisions: simple decisions and multiple-choice decisions (see Figure 3). A simple decision has one incoming branch with one input and two outgoing branches with one output each. When the process is running, the process flow takes one outgoing branch if a certain condition is true and the other branch if the same condition is false. A multiple-choice decision has one incoming branch and multiple outgoing branches. Each outgoing branch has an associated condition (that is, an expression that evaluates to either true or false). This condition determines which branch will be selected when the process runs.

Figure 3. (Click for larger image.)

Classifiers
Classifiers categorize process elements for analysis purposes. Pre-defined classifiers are provided or you can define your own. Colors can be associated with elements that have specific classifier values, enabling you to see these values at a glance (see Figure 4). You can perform static analysis to see the association between classifier values and model elements in the process. After running a simulation, you can perform dynamic analysis to find the total cost or duration of elements with a specific classifier value.

Figure 4. (Click for larger image.)

Data Modeling
Every process model needs a corresponding data model. Data models include the business items, which are the documents, work products, or commodities that are used in the process. Examples of business items are manufacturing orders, mother boards, power supplies, and memory chips (in a PC assembly process), itinerary and customer information records (in a trip reservation process), and passengers (in a transportation process).

Business items are used within a process to model the objects that are acted on by the tasks and decisions. In addition, you can model a specific instance of a business item. For example, if you have defined a business item called Cash Register, you can model individual cash registers (such as Cash Register #7) as business item instances.

To model the flow of data from task to task, you associate business items or business item instances with connections between elements in the process model (see Figure 5).

Figure 5. (Click for larger image.)

Resource Modeling
A resource model is required to accurately model business processes. Resource models represent the people, equipment, or material used to perform a project or a task. Resources are not the same as business items. The objects that undergo changes and are passed from one process step to the next should be modeled as business items, whereas the things that are performing the work or are required prerequisites, such as machines, fuel, vehicles, or skilled personnel, should be modeled as resources.

Each resource is a particular occurrence or example of a resource definition. A resource definition is a set of similar characteristics shared by resources of the same type. If you have modeled a resource definition called Computer, then some examples of resources might be "Main server," "Web server," and "Test machine 1."

Roles, timetables, and costs add additional characteristics to resources. For example, an Employee resource could have the role of Customer Service Representative, Supervisor, or Manager. The Employee resource may only be available on a certain timetable (say, Monday to Friday), and each unit of the Employee's time costs $11.50 per hour (see Figure 6).

Figure 6. (Click for larger image.)

Intelligent Analysis
Now that you've setup a model, its time to delve into WebSphere Business Modeler's analysis capabilities. The tool provides a variety of analysis functions relating to models in their static state. These analysis functions are in addition to dynamic analysis functions, which return information based on the results of running a process simulation.

There are simply too many static analysis tools to describe in this space. Some examples are: analyzing activities unable to start (due to unavailable resources, empty input criteria, or inputs that lack connections); analyzing activity resource and role leveling (shows the allocated and available roles and resources for each activity in a process); analyzing activity cost and duration (shows the cost and duration of cost of each activity in a process); and analyzing qualified resources for a role (shows the resources that are qualified to perform a role defined in the model).

Simulate the Future
A process simulation is a simulation of a real world business process. The business process may be modeled on an existing business process ("as-is") or one that is being planned for the future ("to-be"). Simulation allows you to assess the performance of the process, generate statistics about its execution, and pinpoint potential areas of improvement and optimization.

Simulation enables organizations to observe how a process will perform in response to variations of inputs to the process, just as in a real-life work environment. Simulation also provides the ability to vary process input volume over time by adjusting resources and current allocations.

When you run a simulation, you can watch an animated view of the process in operation (see Figure 7). The animation shows the movement of tokens from the inputs of the process and between activities in the process. A token represents a unit of work that is received by a process and transferred between different activities in the process flow.

Figure 7. (Click for larger image.)

Dynamic analysis lets you extract targeted information based on the results of your process simulations. WebSphere Business Modeler supports four primary types of dynamic analyses:

• Aggregated - Determines information about activities and resources used in all process instances generated during a simulation.
• Process instance - Performs a summary analysis to show process results for activities within a particular instance of a process that is created during a simulation run.
• Process cases analysis - Shows statistics produced by all process cases in a simulation.
• Process comparison - Compares the weighted average analysis results for two simulated processes that use the same input parameters.

Each type of dynamic analysis offers numerous reports that can be generated at the click of a button.

What's Next?
When you're satisfied with the business process model, it is often turned into a software application. Implementation starts by exporting the process definition to a Web Services Business Process Execution Language (WS-BPEL) file, and then specifying key performance indicators that should be used to monitor the process once it is in production.

The WS-BPEL file can then be turned into a UML model using Rational Software Architect. Developers can then generate Java components directly from the UML, which integration developers can integrate with new and existing applications using WebSphere Integration Developer. After deployment managers and business analysts can use WebSphere Business Monitor to track the performance of business processes. So there you have it. WebSphere Business Modeler is the crucial first step in business process automation. It lets you build detailed models of complex business processes with ease. These models can be optimized through analysis and simulation, and eventually turned into software applications by making use of the WebSphere family of business integration tools.