
ment this interface in order to be ModelBus conform. ModelBus itself is delivered
with a built-in model repository, which supports versioning, partial check-out of
models and coordinates the merging of model versions and model fragments.
In a model-driven development process a number of recurrent tasks must be per-
formed like model transformation or model verification. To increase the efficiency of
the underlying development process it is vital to reduce the amount of manual devel-
oper intervention needed and to introduce as much automation as possible in particu-
lar for long running tasks like model simulation. For this purpose ModelBus allows
the definition of single tasks as modeling services and to orchestrate them together
with other modeling services. These orchestrations can automatically be triggered
either by user commands or by other orchestrations.
One of the major problems when dealing with complex models is to keep track of
all incoming and outgoing references (e.g. profiled UML models or diagram models).
As models grow larger and larger it becomes more and more difficult to coordinate
the workflow in a way that permits concurrent work on different parts of a model.
The built-in ModelBus model repository addresses this challenge. Moreover, the
ModelBus notification system can be used to stay informed about the changes applied
to a model by other developers.
Currently a number of tool adapters are already available. The set of ModelBus-
connected tools will be extended by developing new corresponding ModelBus adap-
ters. The organization, the execution and the control of the development process via
BPNM [16] has been implemented utilizing Intalio [17]. Through the adaptation of
Adaptive Repository [18] an alternative repository implementation has become avail-
able.
3 Example Development Process
The ModelBus technology outlined in the section above has been used for the realiza-
tion of a distributed development scenario, which has been applied by partners in the
Modelplex project. In this process several hundred developers of different develop-
ment teams working in different locations are collaborating (see Figure 2). Specific
requirements had to be obeyed regarding access control and model scalability implied
by the complexity of the models involved in this scenario. Furthermore, certain activi-
ties of the development process have had to be automated. This has been realized
using orchestration and notification provided by ModelBus.
In the example development process two development tools are used; Intalio De-
signer for BPMN modeling and Rational Software Architect (RSA) for UML model-
ing. In addition to this, a BPMN-to-UML transformation based on ATL is used as
well as model verification based on OCL, to verify BPMN models according to de-
sign guidelines like naming conventions. All models, development artifacts, model
transformations, and verification rules are stored within the ModelBus repository. For
the automation of the verification and transformation a COTS tools is used. The or-
chestration is deployed as BPEL to an application server in which the ModelBus
repository, the verification, and the ATL transformation engine are running (in Fig.2.
see Node Paris 2).
37