Chapter 1: Overview of Workflow Management

1

• Chapter 1 Overview of Workflow Management

Dr. Shiyong Lu Shiyong_at_wayne.edu Department of
Computer Science Wayne State University
2

• Overview of Workflow Management

Dr. Shiyong Lu Shiyong_at_wayne.edu Department of
Computer Science Wayne State University
3
The DUS Triangle (C. Zhai, 2008)

• Many computing paradigms can be characterized as
  an instance of the DUS Triangle.
• Workflows is one such computing paradigm (a
  workflow connects tasks, agents, data together
  with flows.)

4
What is a workflow?

• A computerized model of a business or scientific
  process consisting of multiple workflow tasks and
  their coordination dependencies.
• Workflows are essential to enterprise operation
  as most business activities of an enterprise
  today are computerized and automated using
  workflows.
• Workflows become increasingly important for
  scientific computing as more and more scientific
  computation and analysis processes are
  represented and executed as workflows.

5
A workflow example
(S. Lu, 2002)
6
What is a Workflow Management System?

• A Workflow Management System (WFMS) is a system
  that supports the specification, execution, and
  monitoring of a workflow.
• Supports a high-level workflow specification
  language (e.g., BPEL)
• Applications or users describe their workflows
  using that language.
• WFMS interprets a workflow of that language by a
  coordinated execution of the tasks in the
  workflow.

7
Major components of a WFMS
(C. Lin, et al. 2008)
(Hollingsworth, 1995)
8
Process definition tool

• Defines a workflow using a workflow specification
  language in a textual, graphical, or scriptual
  format.
• The output of such a tool is a workflow
  definition that typically includes
• Workflow name, version, creation time, author,
  description, keywords, etc
• Workflow tasks
• Coordination dependencies, typically in terms of
  controlflows and dataflows.

9
Workflow enactment service

• Interprets a workflow specification by creating,
  executing, and managing runtime instances of
  workflows
• Consists of one or more workflow engines
• Schedules the invocations of workflow tasks
• Maintains the status of runtime instances of
  workflows and tasks.

10
Workflow client applications

• Supports the execution of workflow tasks that
  need to be performed by humans (human tasks).
• A workflow engine creates and enqueues work items
  to a worklist.
• A worklist handler dequeues and routes work items
  to different users for performing.
• Issues communication mechanism, access control,
  scheduling, load balancing, user interface design.

11
Invoked applications

• Supports the execution of workflow tasks that are
  performed automatically by computers (automatic
  tasks)
• Invoke various local or remote heterogeneous
  applications
• Session, data, execution, and event management.

12
Interoperability

• On one hand, a workflow might go across multiple
  enterprises, on the other hand, different
  enterprises might choose different WFMSs.
• Interoperability at various levels
• Subsystem
• Task
• Workflow
• An effort led by the Workflow Management
  Coalition (wfmc.org)

13
Administration and monitoring tools

• Administration
• User management
• Role management
• Audit management
• Process supervisory functions
• Resource control
• Monitoring
• Status, progress, performance of execution of
  workflows
• Failure report and human intervention

14
What is a Workflow Application System (WFAS)?

• An ad hoc system that is deployed based on a
  particular workflow specification typically with
  the support of a WFMS.
• It only supports one or a set of fixed closely
  related workflows for a particular workflow
  application.
• Workflow evolution is possible but usually needs
  to go through a formal business re-engineering
  procedure.

15
A typical deployment of a WFAS
16
System requirements

• Availability. It is important for an enterprise
  to conduct its normal business operations.
• Reliability. Workflows need to be executed
  reliably even in the presence of failures and
  concurrency. Status and data cannot be lost.
• High throughput. Efficiency of a companys
  business operations. Many users and many task
  runs/per second.
• Scalability. The system should scale well as the
  demands and resources increase.

17
System requirements (cont)

• Security. As the system is accessible by many
  users, potentially from different geographically
  distributed enterprises, security is important.
• Interoperability. On one hand, a workflow might
  go across multiple enterprises, on the other
  hand, different enterprises might choose
  different WFMSs.

18
Developers of a WFAS

• System analyst. The system analyst works with
  the customer to identify business rules,
  requirements, and policies and produce a
  requirement analysis document.
• Workflow engineer. The workflow engineer designs
  a workflow specification based on the requirement
  analysis document.
• Application programmer. Implements individual
  workflow tasks and interfaces of the workflow and
  deploys the system.
• Project manager. Direct and oversee the
  successful completion of the whole project.

19
Users of a WFAS

• End users. Perform human tasks assigned to them.
• Workflow administrator. 1) Manage workflow end
  users and applications 2) Monitors workflow
  execution and 3) Intervenes during failures of
  workflow execution.
• System administrator. Responsible for the whole
  system 1) hardware and software need, 2)
  configuration, and 3) performance and security.

20
The Trend of workflow
(1/10/2012)
http//www.google.com/trends, interpret it on
your own risk!
21
Workflow vs. software engineering
22
Workflow vs. bioinformatics