Traditional information technologies focus on supporting specific functions rather than supporting the
overall underlying work procedure (so-called business process). Workflow management systems allow
the execution and coordination of such business processes (also called workflows) in an integrative
manner.
A workflow management system typically consists of two components. The modelling component
provides functionality for specifying the abstract structure of a workflow. On the other hand the
enactment component is responsible for the real time management of the steps of a business process.
Therefore the enactment component must be aware of relevant changes concerning the actual
processing of a process, i. e. after completing a piece of work by an agent the system has to proceed
with the further actions: it has to determine the following pieces of work, it has to establish
corresponding agents for those pieces of work and it has to deliver the necessary information to those
agents.
These requirements can be well served by active databases which extend the functionality of
conventional ("passive") databases with a rule mechanism. The relevant events and conditions are
specified by rules to the active database. This enables the active database system to monitor the actual
state of the database and to invoke actions automatically if the specified event occurs.
To show how this rule mechanism can be used for implementing the enactment component efficiently a
prototype workflow system (called PANTARHEI) has been developed. PANTARHEI introduces a special
purpose description language for specifying business processes (called WDL: workflow description
language). Specifications done with WDL can then be transformed automatically in a set of rules which
provides the required functionality. After creating these rules in the active database, the system can
coordinate and schedule the specified processes directly.
The usage of an active database for implementing a workflow system therefore has a substantial payoff
in reducing the development effort for such a system. Important features of the database system as
distributed processing, recovery, concurrency control and rule mechanism are available immediately and
need not be reimplemented.