Post on 16-Dec-2015
Anatomy of a Workflow Engine
What is a Workflow Engine? Orchestrates business process workflow It accomplishes this via:
routing content through a variety of routing nodes
What must it be capable of doing? Modeling both simple and complex processes Interacting with multiple entities (both
humans and services) Running asynchronously Transactional operation
KEW Overview
Composed of a set of services APIs and User Interfaces
Routes documents based on type, where the associated Document Type definition declares a route path for the document.
Based on this route path, various behaviors will be performed based on the document content.
Often, this behavior consists of interacting with one or more parties in the form of approval requests. For example: Transfer of Funds Timesheet Drop Course
KEW – Core Features
Action List (User’s Work List) Document Searching Document Audit Trail (Route Log) Flexible Process Definition (Document Types) Simulation Engine Notes and Attachments – All workflow documents get
them, even if they don’t want them Programmatically extensible Wide array of pluggable components to customize
routing and other pieces of the system
Document Types
Document Types allow you to specify content that can be initiated using KEW
They are specified using XML markup They identify this content by a type name Can also include policies and other
attributes Hierarchical, can inherit attributes from
parents Most importantly, they define the route
path for initiated content
Process Definition - Route Nodes
Some of the primary types of Routing Nodes include: Split Nodes Join Nodes Simple Nodes Request Nodes Dynamic Nodes Sub Process Nodes
You can also implement custom Nodes
Route Nodes - Split
A split node is one that is capable of splitting the routing path into multiple branches
Branches are executed concurrently
Route Nodes - Join
Join nodes complement a split node by allowing the multiple branches of the routing path to converge
Route Nodes – Simple Node
Simple nodes are a base type that allows a developer to implement behaviors based on their own particular processing requirements
Simple Node Examples
EmailNode: Allows sending of email messages
LogNode: Sends messages to Rice’s logging stream
SetVarNode: Allows setting of document variables
RequestsNode: Generates action requests
Route Nodes - Request Nodes
Generates ActionRequests that are fed into the action list(s) of a user, workgroup, or role
Requires user action to proceed Include: approve, disapprove, acknowledge,
etc… Have an associated Activation Type:
S (Sequential): An ActionRequest is sent and must be acted upon before sending the next ActionRequest
P (Parallel): Multiple ActionRequests are sent at once Requests can be generated by a custom Route
Module or by the Rules engine
Route Nodes – Dynamic Nodes
Dynamic nodes are a more advanced form of developer-implemented node
They allow for the dynamic modeling of a route path
This is done by instantiating a set of node instances that will then be fed to the routing engine
Route Nodes – Sub-Process Sub Process Nodes are named route
paths that can be included in a document type’s primary route path.
They allow the workflow author to create a route path that can be reused
Custom Node Implementation
Making an example of the Email Node:
public class EmailNode implements SimpleNode {
public SimpleResult process(RouteContext context, RouteHelper helper) {
if (context.isSimulation()) {
// no side-effects!
}
else {
// send off the email
}
return new SimpleResult(true);
}
}
KEW API
Consists of two main services:
1. WorkflowDocumentActions All operations related to a workflow
document
2. WorkflowUtility Implements various query operations
WorkflowDocumentActions
Create Route Approve Save Blanket Approve Return to Previous Node Super User Actions Attach XML to document Set document title
WorkflowUtility
getRouteHeader getDocumentType getAllActionItems isUserInRouteLog routingReport (executes the Simulation
Engine) performDocumentSearch ruleReport getActiveNodeInstances getDocumentStatus
Rule Routing
Four key components used for Rule Routing Document Type (specifically the Route
Nodes) Rule Attributes Rule Templates Routing Rules
Document Type (Route Nodes)
Sample Route Node using ‘Rule Template’ Method
<requests name="DocumentNode">
…
<ruleTemplate>DocumentRuleTemplate</ruleTemplate>
</requests>
Sample Route Node using ‘Named Rule’ Method<requests name="DocumentNode">
…
<ruleSelector>Named</ruleSelector>
<ruleName>TestRule1</ruleName>
</requests>
Rule Attributes
Defines how a Routing Rule evaluates document data to determine whether or not the rule should fire.
Also defines what data is collected on a rule.
An attribute typically corresponds to one piece of data on a document (i.e. dollar amount, department, organization, account, etc.)
Can be written in Java or defined using XML
Rule Templates
A composition of Rule Attributes When a Routing Rule is created, it is
created for a specific Rule Template and Document Type.
Defines what data should be collected and stored in the rule (via it’s Rule Attributes)
Also used to identify the set of rules to evaluate at a “Requests” Node
Can also be used to define various other rule properties (defaults, etc.)
Routing Rules
Evaluated against a document to determine who should receive action requests at a particular node.
Created against a Rule Template. Rule matching behavior is handled by the
Rule Attributes when the rule is executed. Defines the principals (users), groups
and/or roles who should receive action requests should the rule fire.
Examples: If dollar amount is greater than $10,000 then
send an Approval request to Joe. If department is “HR” request an
Acknowledgment from the HR.Acknowledgers workgroup.
KIM Responsibility Routing New to Rice 1.0.x Leverages Kuali Identity Management
module to allow routing to KIM Roles via KIM Responsibility objects
Allows for both KIM Permissions and workflow routing to use same KIM Roles
Can be used with ‘Rule Routing’ on a single document type but not single node
Document Type Configuration
Node must be defined using <role> xml tag rather than <requests> tag used for ‘Rule Routing'
<role> tag identifies to KEW engine that node will use KIM Responsibility Routing
<role> xml tag is used in <routePath> and <routeNodes> elements
Qualifier Resolvers
Used to find qualification data for matching to KIM Role Members
Data can be located anywhere including external systems
Defined on Node using fully qualified class name
Single class can be used on multiple document types and multiple nodes
Qualifier Resolvers (cont)
Can be custom coded java class implementing QualifierResolver interface
Some provided options included in Rice NullQualifierResolver (Match All Role
Members) XPathQualifierResolver (User provided
XPath Expression) DataDictionaryQualifierResolver (KNS
Specific)
KIM ‘Responsibility’ Business Object
KIM specific object Very similar to KIM Permission object Usually one or more KIM Roles are
assigned to a Responsibility Each Responsibility has a Namespace
value (KR-SYS, KFS-AR, KS-LUM) but these are not used by KEW routing
Responsibility has data values associated with it via ‘Detail Value’ elements
Responsibility Detail Values KEW uses specific Responsibility Detail
Values to find matching Responsibility objects Document Type Name Route Node Name
Responsibility create/edit UI is designed to collect this information
Can be viewed from the Responsibility Lookup
Responsibility Detail Values (cont)
documentTypeName – The KEW document type name that the responsibility applies to
routeNodeName – Specific Node Name to use this particular Responsibility for
actionDetailsAtRoleMemberLevel – Tells KEW where to find Action Info (request type, priority, etc)
required – Notifies KEW if this Responsibility is required to generate at least one Action Request
KIM Roles
Qualification values returned by node’s Qualifier Resolver class are used to fully qualify Role Members
The Responsibility Detail Value ‘actionDetailsAtRoleMemberLevel’ determines how Role collects Responsibility ‘Action Info’ True: Each Role Member will have it’s own
‘Action Info’ data False: The ‘Action Info’ data will be set at the
Role Level and every Role Member uses it
Anatomy of a Document Type Definition
<documentType> <name>...</name> <parent>...</parent> <description>...</description> <label>...</label> <postProcessorName>...</postProcessorName> <docHandler>...</docHandler> <superUserGroupName namespace="...">...</superUserGroupName> <blanketApproveGroupName namespace="...">...</blanketApproveGroupName> <defaultExceptionGroupName namespace="...">...</defaultExceptionGroupName> <security>...</security> <notificationFromAddress>...</notificationFromAddress> <serviceNamespace>...</serviceNamespace> <attributes>...</attributes> <policies> <policy> <name>...</name> <value>...</value> </policy> </policies> <routePaths> <routePath> ... </routePath> <routePath name="..."> ... </routePath> </routePaths> <routeNode> ... </routeNode></documentType>
Defining Route Paths
<!-- This section defines the path of the nodes or the "flow" --><routePaths> <routePath> <start name="Start" nextNode="DepartmentChair" /> <role name="DepartmentChair" nextNode="FinalApprovers" /> <requests name="FinalApprovers" nextNode="ConfirmationEmail" /> <simple name="ConfirmationEmail"/> </routePath></routePaths>
<!-- This section is used to configure each node --><routeNodes> <start name="Start"> <activationType>P</activationType> </start> ...</routeNodes>
Designing Workflow Processes
Next, we’re going to take a look at some Document Type definitions in XML.
We will look at some of the Document Type definitions you will work with in the next exercise: Simple Document Type with Ad-Hoc Routing Routing Rules using Simple Rule Templates Parallel Branching using Split Nodes Using Conditional Split Nodes Using a PostProcessor
Different Types of Nodes
<routePaths> <routePath> <start name="Start" nextNode="DepartmentChair" /> <role name="DepartmentChair" nextNode="FinalApprovers" /> <requests name="FinalApprovers" nextNode="ConfirmationEmail" /> <simple name="ConfirmationEmail"/> </routePath></routePaths>
<routeNodes> <start name="Start"> <activationType>P</activationType> </start> <role name="DepartmentChair"> <activationType>P</activationType> <qualifierResolver>DepartmentCodeQualifierResolver</qualifierResolver> </role> <requests name="FinalApprovers"> <activationType>P</activationType> <ruleTemplate>FinalApproversTemplate</ruleTemplate> </requests> <simple name="ConfirmationEmail"> <from>myemail@localhost</from> <to>initiator</to> <testAddress>testemail@localhost</testAddress> <style>ConfirmationEmail.Stylesheet</style> <type>edu.iu.uis.eden.mail.EmailNode</type> </simple></routeNodes>