Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Parallel Routing Mechanism in Document Routing

IP.com Disclosure Number: IPCOM000126683D
Original Publication Date: 2005-Jul-28
Included in the Prior Art Database: 2005-Jul-28
Document File: 7 page(s) / 42K

Publishing Venue

IBM

Abstract

Content Manager expands on the workflow function offered in prior releases. One enhancement is the addition of Parallel Routing functionality. This type of work node (which facilitates parallel routes) is modelled by the CM graphical Builder as split and join routes. While parallel routing is a common feature of workflow systems, the modelling and implementation within Content Manager to achieve the best possible usability, performance, and function requires an innovative design. Specifically, a method is required in which a separate, cloned work package is created for each parallel route, but the data associated with the work package (workflow variables, priority, assigned user, etc) are kept identical for all instances of the work packages. The main idea is that the original workpackage will stay in the split node when the cloned workpackages are routed through the parallel routes. Whenever a cloned workpacakge reaches the join node, it will stay in the join node waiting for all other cloned work packages to arrive at the join node. When all the cloned workpackages arrive at the join node, the cloned workpackages will be destroyed and the original workpackage will be moved from the split node, where it stays when the cloned ones are routed, to the node right after the join node. By supporting nested parallel routing (as described in section 2), CM parallel routing provides full capability to model almost all kinds of real world parallel routing scenarios.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

Page 1 of 7

Parallel Routing Mechanism in Document Routing

One document traverses multiple routes in parallel routing. It can be at multiple work nodes simultaneously. A "Split" node indicates the beginning of parallel routes and a "Join" node waits for completion of all routes. Split and join nodes come in pairs. A pair of split and join nodes encompassing multiple parallel routes can reside on a parallel route in the scope of another pair of split and join nodes.

Content Manager (CM) parallel routing provides a unique solution to the parallel routing problem in workflow. Each parallel routing process consists of a split node and a join node, as shown below, where WN1, WN2, etc represent regular worknodes in the workflow process.

WN1 --> WN11

/ \

SPLIT_NODE --> WN2 --> WN12 --> JOIN_NODE

\ /

WN3 --> WN13

A workflow process may have more than one parallel routing process:

WN1 --> WN11

/ \

START_NODE --> WN0 --> SPLIT_NODE1 --> WN2 --> WN12 --> JOIN_NODE1 -> WN4 ->

\ /

WN3 --> WN13

WN5 --> WN51

/ \

SPLIT_NODE2 --> WN6 --> WN61 --> JOIN_NDOE2 -> WN8 -> STOP_NODE

\ /

WN7 --> WN71

Inside a parallel routing process, there could be nested parallel routing processes. CM parallel routing imposes no restriction on the number or level of nested parallel routing processes, providing full capability to model any real-world workflow scenarios.

WN1 --> WN11 -------------------------------------------------\

/ \

/ \

/ \
SPLIT_NODE --> WN2 --> WN12 ----------------------------------------------------->
JOIN_NODE

\ /

\ /

\ WN131 --> WN134 /

\ / \ /

WN3 --> WN13 -> SPLIT_NODE1 --> WN132 --> WN135 --> JOIN_NODE1/

\ /

WN133 --> WN136

One of the issues to be solved in the parallel routing is to keep the work flow data in synch while going through parallel routing. In CM document routing, a workpacakge keeps a "pointer" to the real document to be worked on. The real document is not cloned for the cloned workpackages created during parallel routing. When a cloned workpackage is retrieved, the real document is "de-referenced" from workpackage. To ensure data integrity of the document, the CM check-in/check-out mechanism is used so that at any moment, only one user can work on the same document.

In the beginning of the parallel routing, the priority, assigned user, workflow variables (container data) are

1

Page 2 of 7

cloned for all the cloned workpackages. During parallel routing, priority and assign user in the cloned workpackages may be changed in the parallel routes. There is no need to keep them the same for all the cloned workpackages.

By contrast, it is desirable to keep the workflow variables or the container data in synch during the parallel routing. Whenever a new workflow variable, say loan amount= $100,000, is added to a cloned workpackage during the parallel routing, this workflow variable will be cloned for all other cloned workpackages and the original workpackage. In so doing, any change in the workflow variable will be immediately made visible to all...