Open a Workflow Task or Instance Trail from Oracle BAM

An interesting requirement that I have recently seen a lot of clients demanding in their BAM dashboards is the ability to action or monitor process instances. Oracle BAM 11g typically provide means to create real time dashboard for process metrics to business process stakeholders. However a lot of organizations are beginning to prefer to also use it as a starting point to get more insight into instance details (such as in-flight status, flow trail, fault information, etc.). One of my reputing requirement in the past was so much as to provide a mechanism to open actionable task forms directly from BAM. And then sometime back I had the same question asked again as how to open a process instance trail directly from Oracle BAM 11g to view its fault details. This is a natural demand for process owners typically in the support role who are alerted of fault notifications from BAM and would not like to go through the trouble of going back and forth the BAM dashboard and the Enterprise Manager console to correlate instances together. This eventually led me to believe that, it is in fact a sought after requirement and hence this article will aim to provide a detailed and step by step guide to define and implement this use case.

In a nutshell, this article will demonstrate creation of actionable links in a BAM dashboard to view and instance trail of a composite instance and from there dig into its audit trail or flow trace. Another useful feature demonstrated will be the use of Action Buttons in Oracle BAM 11g dashboard to launch instances of human workflow tasks of a BPMN composites awaiting to be actioned from designated process users. When such an instance is clicked the user is taken to the BPM workspace UI and take an action on the selected task.

This allows business users to seamlessly monitor business applications and processes and be able to correlate them with instances appearing across different UI’s of the Oracle SOA Suite 11g product suite such as BPM Workspace and EM Fusion Middleware Control console. It is essentially useful for organizations that intend to use Oracle BAM as an operational dashboard for process users in addition to it being used as a real time business metrics monitoring and alerting dashboard.

Take a Sneak Peak at the Use Case

The following video will show a sneak peak of the use case intended to be demonstrated in the course of this article.

During the course of implementing the use case, the article will also describe the following activities in detail:

• An Overview of creating and assigning values to a business indicator and publishing it to Oracle BAM data objects at runtime.

• Configuration changes involved in sending the process metrics data from a BPM process to BAM data objects such as activating sampling points, enabling data targets, configuring the OracleBAMAdapter and altering BPMN engine MBeans .

• Create Streaming Action Lists in an Oracle BAM reporting view to track in-flight instances of a composite that are assigned to process owners, waiting to be actioned. and a list of completed instances.

• Create an action list of completed instances and define an action to launch their audit trail in the Enterprise Manager console. This concept is useful as this can be extended/customized to instead create a list of faulted instances for users to gain detailed insight into the fault details from the Enterprise Manager Fusion Middleware Console.

• Configure External Data Objects and create lookup fields in BAM.

• Create and configure action buttons in Oracle BAM to open an external URL and pass the value(s) of a column in an action list for a selected instance to the URL term.

To gather process metrics and activity instance data in BAM to be able to build the reports, an existing BPMN based composite will be used. The composite project can be downloaded from the example demonstrated in an earlier BPM post on event based gateways. The article also describes the use case the project tries to address in details.

 

https://beatechnologies.wordpress.com/2012/05/12/understanding-event-based-gateways-in-oracle-bpm-suite-11g/

The download link for the application workspace is provided in the aforementioned article. The soapui project to create sample instances and thus populate the BAM reports is provided at the end of this article.

Prerequisite(s)

The modelling workspace and the version of Oracle BPM and BAM Suite used in this demonstration is 11gR1PS5 i.e 11.1.1.6 although the concepts discussed are generic and applicable to all versions.

 

Creating the Business Indicator(s)

Import the application workbench consisting of the modelled process in JDeveloper. To begin with the article describes adding a business indicator to capture the credit card approval status as the instance progresses through multiple activities.

In the JDeveloper menu click on View > BPM Project Navigator and open the AccountOpeningProcess business process. You will notice a Structure pane visible in the bottom left hand side of the editor. If the structure pane is missing then click on View > Structure to be able to see it. Expand Business Indicators > Dimension and right click on it to be able to add a new dimension of type String. Name the dimension as cardApplicationStatus. For simplicity this article just has a single business indicator, though in real business processes you may end up having multiple KPI collecting indicators.

Once an indicator has been defined, it has to be assigned a certain value to depict a business friendly state of the process instance is currently in. Have a look at the red information rectangles for the AccountOpeningProcess as depicted in the screenshot below. They may well represent the various business friendly stages of the application status. For example, when a New Application request is received  by the process the business indicator may be set to “PPLICATION IN PROGRESS indicating that it is being currently processed upon. If the customer for any reasons wants to terminate his application midway while his initial request is in-flight, the application status may be transitioned to “REQUESTED CANCELLATION”. Similarly there may be other intermediate statuses that can be captured in a long running business processes and be presented to business stakeholders or process owners.

Populating a business indicator with a pre-determined value is straightforward. It is in a similar way as a value is assigned to any other process variable. Double click on the New Application start message event and then on the Data Associations link. Simply drag the Expression icon on the cardApplicationStatus metric variable under OnlineAccountOpenin>DataObjects on the right hand side of the Data Association wizard and assign it a static value of “APPLICATION IN PROGRESS”.

Similarly, double click the Cancel Application intermediate event and then assign REQUEST FOR CANCELLATION to the same business indicator from the Data Associations link. When the business processes receives an intermediate request for cancellation while the task is still in progress, the cancel application intermediate event is invoked and the latest status of the process instance (determined by the cardApplicationStatus business indicator) is replaced. Use the same technique to assign other possible approval states to the business indicator for other activities.

Configuring Sampling Points and Data Targets

Now that the business indicators are assigned appropriate values describing meaningful process states, the next thing is to publish these indicators to BAM. Readers familiar with Oracle SOA Suite will know that sensors can be defined for the BPEL engine to automatically publish their data to BAM data objects in real time as soon as their values in the business process changes. Unfortunately as of now, sensors cannot be defined in Oracle BPM Suite. Instead there is a concept of Sampling points.  The BPMN Service Engine uses the defined sampling point configuration to decide when and where to publish process analytics information. Process analytics sampling points can be created for all activities, just interactive activities or disabled altogether. The process metrics can be published internally to the in-built cube star tables in the SOAINFRA schema or externally to BAM data objects. The process sampling points can be configured by right clicking on the BPM process from the BPM Project Navigator view and then clicking Project Preferences . This article needs sampling points to be generated for all activities in the business process, hence Generate for All activities from the Project Sampling Points radio options is selected. Sampling points capture a host of other information in addition to capturing the values of defined business indicators such as performance and workload metrics. Also note that choosing Generate for All Activities will generate a lot of data to be published to BAM. More information about Process Analytics and Sampling points can be read at the official Oracle documentation site at

http://docs.oracle.com/cd/E28271_01/doc.1111/e15176/monitoring_proc_act_bpmpd.htm

By default the sampling point only publishes process metrics to the internal CUBE_* tables in the SOAINFRA schema. To publish these analytics data externally to BAM click on the Data Targets tab and check Enable BAM option. Also change the default Data Object Path to something like /Customer/AccountInitiation. This ensures that all data objects pertaining to a project are created in the folder hierarchy entered in the data object path field.  This external publishing of the BPMN engine data to BAM data objects in the active data cache is taken care by the OracleBAMAdapter. This adapter has two predefined connection factories. It allows either publishing the data to the BAM engine either through a RMI interface or through a web service interface.  The BAM Adapter JNDI name for these interfaces is either eis/bam/rmi or eis/bam/soap respectively. The eis/bam/rmi provides a better performance as this doesn’t employ the BAM web service interface and used preferably if the BAM and BPMN engine reside in the same JVM.

Configuring the OracleBAMAdapter

The integration glue that binds and permeates instance metrics information to pass from the BPMN engine to the BAM active data cache is the OracleBAMAdapter. The  BAM adapter needs to have the same JNDI (one that was selected in the previous step) configured that is pointing to the correct instance of the BAM active data server. Oracle Weblogic Server administration server console provides the mechanism to view and edit BAM adapter configuration properties. Login to the administration and navigate to Summary of deployments > OracleBamAdapter > Configuration > Outbound Connection Pools and click on the eis/bam/rmi connection pool to view its default configuration properties. These properties may have to be changed to point to the correct instance of the BAM active data cache server. It is also important to note that any change in these adapter properties are applicable only when they are saved and the adapter configuration is updated. This typically happens when the adapter is either redeployed with the changes in the plan file or updated with  the saved changes.  Proper configuration of all of the above settings is important for the BPMN engine to be able to plug with the BAM server.

On an Oracle BPM Suite 11g PS3 and above, with only these changes in place, simply deploying the BPM project will create the necessary data objects in the BAM active data cache. Open the Oracle BAM console and click on the BAM Architect view to be able to see the list of data objects existing in the BAM active data cache server. Notice the AccountInitiation subfolder created under DataObjects > Customer folder. Under this hierarchy will be the data objects containing the instance metrics data published from the BPMN engine. Later parts of this article will show how these data objects will be used to create interactive reporting view in BAM.

BPMN Engine MBean configuration Alteration to Allow the engine to post data to BAM

As discussed earlier, the above set of configuration changes are sufficient to create the data objects when the business process composite is deployed to the application server. Now when an instance is created and executed it is expected that these data objects will have process metrics value populated at near real time. However the contents of the data objects are not be populated unless a certain disable action to supress process measurements to be sent to the BAM engine is turned off. Login to Oracle Enterprise Manager Fusion Middleware console and navigate to soa-infra > SOA Administration > BPMN Properties > More BPMN Configuration Properties to view the configurable BPMN engine MBean properties. Scroll down the list to see the DisableActions property. The default value for this property is BAMCommand, which supresses measurement actions to be sent to BAM. Clear this field (leaving it blank) and save the settings. If another instance of the BPMN process is now executed, the BAM data objects will being to be populated with measurements obtained for the  various process activities.

Creating the Reporting Dashboard in BAM

With the data objects being able to receive measurement data from the business processes, the next thing to do would be to create the BAM reporting dashboard. The dashboard to be developed in this article will have two horizontal regions each displaying a streaming action list of its own. The idea is to have one of the horizontal view displaying recently completed instances and have a hyperlink defined to launch its detailed audit trail in a pop-up while displaying the list of in-flight instances in another. The view showing the in-flight instances will also provide an action button to launch the approval task form for the the selected instance.  As a desirable feature a similar  streaming list of displaying the faulted instances can also be created that leads process users enlisted as support stakeholders to the flow trail to know the exact point and reason of failure.

Login to the Oracle BAM 11g console and from the landing page click on Active Studio (this is where reports and alerts are create in BAM). Click on the Create a New Report link to open the tiled report template selection screen. Select the template titled Two equal horizontal tiles with thin separator. From the list of various report types in each of the horizontal tiles select Action List and  title the report as Credit Card Approval Scoreboard. Each of the report views will prompt for a data object to be selected from which the view will sources its data content.

The upper horizontal tile is used to display a list of completed instances. For this view select BI_default_OnlineAccountOpening_AccountOpeningProcess as the data object and then click on the Fields tab. This lists all the fields in the data object. From a reporting standpoint, it may not necessarily make sense to display all the fields in the data objects but only a few important ones.  For the report in this article, the selected fields are COMPOSITE_INSTANCE_ID,  BI_NAME, COMPONENT_START_TIME, COMPONENT_END_TIME, COMPONENT_INSTNANCE_STATUS, COMPOSITE_NAME, COMPOSITE_REVISION, DOMAIN_NAME and METRIC_cardApplicationStatus. These fields however will appear in the report in their alphabetical order. To change this check the select-boxes corresponding to these fields and use the Arrange icon to rearrange them to be displayed in a custom order. Optionally the report can also be pre-sorted based on values a particular field by moving the field under the Sorted Fields panel in the Sort tab.

Another important thing worth mentioning here that if the view configuration is left just as it, a whole lot of duplicate instances will appear in the list. They are not necessarily duplicate as the process measurement data is published for every activity. Therefore the same composite instance will have as many rows in the data object as the number of activities in the original business process. There will however be only one row with the latest snapshot of the process. In order to show only the unique and non duplicated list of composite instances in the streaming list the Filter tab can be used to and add a filter expression to discard the non-latest snapshot of an instance. Add a new filter entry and for the filter expression select the field as LATEST from the data object fields, the comparison type as is equal to and equate it to a value of Y. For the report in this article, yet another filter is added to display only the approved instances by adding a filter expression to equate the METRIC_cardApplicationStatus field to a value of APPROVED. The list can be narrowed further by adding more filters to the list.

The Properties tab allows to change the view name, fonts and other visual aspect of the streaming list.The final streaming list view should resemble something as shown in the below screenshot.

Hyperlinking a Column in the Report with an Action

Well, the teaser video above showed that the composite instances in the view were hyperlinked which when clicked pops up the audit trail (as seen from the composite dashboard of Enterprise Manager Fusion Middleware Console) for the selected instance. Now for the creamy part of this article, it will be described to create and implement action buttons in Oracle BAM and how they can be used in a report. Select the Completed Instances streaming list report tile and then click on the Change Report Properties link under the report Actions menu. This opens the Report Properties popup dialog. Click the Actions tab and then the New button to create and configure a new action for the report. Provide a name and optional description for the action name and hit Next.

The Action Type lists various action that can be performed on data in a report such as executing a read/write action on data input, selections from that action list or form, open an external URL etc. From the options in the select the Open a URL radio box and then click on Next to open the Action Editor. The action editor provide a very convenient feature to construct a URL by creating a combination of terms term which can be concatenated in the end to form the final URL. These terms can be configured to either have constant values or represent dynamic columnar values from any selected view in a report.

Before being able to configure the URL term, it is required to know how to open the link to the audit trail of an instance. The audit trail of any composite instance can be directly viewed in the browser by using the link below.

http://<hostname>:<portName>/em/faces/ai/soa/messageFlow?target=/Farm_<domain_name>/<domain_name>/soa_server1/<partition>/<composite_name>%20[<version>]&type=oracle_soa_composite&soaContext=<partition>/<composite_name>!<version>/<instance_id>

The link, however is made up of a lot of instance specific information such as the host, port and partition of the soa server where the composite is deployed to, its version and the instance id. Interestingly all but the host name, port name and the domain name (Weblogic server domain name) can be figured out from the default data object itself. They can either be hard coded for each instance of the BAM server to have the corresponding value of the SOA server or the other ingenious way is to create a table in the database and read the values from it. The instance specific information can be directly derived as values of the selected row from the view list. There are other ingenious ways to do it if you dont want to hard code these values. Entering constant values for a URL term is pretty simple. Simply choose the Mapping Type as Constant Value to be able to enter the hard coded part of the URL.

The dynamic part of the URL value can be added in a two step process. First select the choose Value from a List View radio buttonin the Mapping Type and then select the view from the report skeleton containing the streaming list to populate a drop down of the data object columns used in the list. Select the appropriate column value that is needed in the URL. For instance, the COMPOSITE_NAME, COMPOSITE_REVISION, DOMAIN_NAME (this is the partition contrary to the Weblogic domain name) and COMPOSITE_INSTANCE_ID may be required as dynamic values in the URL. There is no restriction on the number of terms that can be created as a combination of constant and lookup values to form the complete URL.

The final Open a URL configuration screen will contain multiple entries that are a combination of constant hard coded values and dynamic values from the list. These are concatenated together to form the actual URL for the hyperlink to be launched when the composite instance column is clicked in the list view. The wizard also allows editing the features of the popup. Click on the Click here to edit Windows features link and choose Full screen for both the width and height of the popup window to display it in full screen mode.

Now that the action is defined the last remaining bit is to configure the action button itself. Click on Next to be presented to the Button Formatting screen. Select the Display the button in a view radio button and from the visual empty report  skeleton select the streaming list view to which the URL action needs to be added. As a hyperlink is desired on the composite instance id column for the view check the Display as a column of formatted links in the selected view option and from the Column to format dropdown choose COMPOSITE_INSTANCE_ID. Click OK on all the screens to save the changed report properties.  With this the configuration changes required for the Completed Instances view of the reporting dashboard is complete. The article will shortly demonstrate how the final view looks like and what does the configured action list do.

Adding the SOAINFRA Schema as an External Data Source

In the following part of the article, another interesting aspect of the use case that is to view the task form of an instance that is awaiting an action from a process user will be demonstrated. But before going ahead with creating the report view there is something else that has to be taken care of. In order to open a task form corresponding to a particular instance id, its taskId is required. Unfortunately the default business indicator metrics data object does not contain the task id. The easiest way to lookup taskId for a composite instance is to query the WFTASK table in the SOAINFRA schema.

select taskid,state from wftask where compositeinstanceid=?;

However the problem is that composite instance id and taskId share a one to many relationship. The query to retrieve taskdId from compositeInstanceId my return more than one row and not all of them can be used to open the task form. Another important thing to know would be that only the taskId row that has the value of the state column as ASSIGNED indicates the current task(s) that are pending actioning from process users. This is achieved by simply appending a filter on the state column.

select taskid,instanceid,ecid from wftask where compositeinstanceid=? and state=’ASSIGNED’;

It is therefore required that there is some sort of mechanism available in BAM to lookup these required field from the database table. This is facilitated in Oracle BAM using external data sources. To begin with open the BAM Architect view from the landing page and click on External Data Sources. Click Create to configure an external data source, provide a suitable name, say SOA_INFRA and connect to the [PREFIX]_SOAINFRA schema. Check out the screenshot below to see as how to provide the schema name, password and the connection string. Verify that the connection to the schema is established by clicking the Test button at the bottom.

Creating an External Data Object with the WFTASK table

The external data source can now be used to create an external data object. Browse to the AccountInititation subfolder and click on the Create Data Object link to create a new data object. Name it as TaskDetails. Check the option box beside External Data Object  indicating that this data object sources its content from a database table instead. Select the external data source that was created in the previous step and then select WFTASK from the dropdown under External Table Name. The columns of this database table are now available to be added as fields for the data object by clicking on the Add a field link. The fields that are added from database table are COMPOSITEINSTANCEID,TASKID,PROCESSDUEDATE,ACCQUIREDBY,ASSIGNEES,ASSIGNEEUSERS,CREATOR,STATE,INSTANCEID,ECID. Note that only a few of these fields will be needed to configure the report view. After the fields are added click on Save Changes to save the data object. View the Contents tab of this data object to see the selected columns for all the rows in the database.

Adding Lookup Fields in the Business Indicator Data Object from the External Data Object

As discussed earlier the streaming action list view to display the in-flight composite instances with human tasks will need to source its reporting data from the BI_default_OnlineAccountOpening_AccountOpeningProcess data object. But somehow this data object would need the additionally  required fields too such as the TASKID, STATE. These fields can be looked up from the TaskDetails data object. This is straight forward as Oracle BAM data objects allows creating look up fields that can be created to look up data from different data objects. Open the BI_default_OnlineAccountOpening_AccountOpeningProcess data object and click on Layout > Edit Layout to be able to add lookup fields to it. Once in the edit mode, click on the Add one or more lookup fields link at the bottom to open the lookup field definition popup. Select the TaskDetails data object and from the list of Lookup Fields click on the TASKID field. This lookup however will be based on some common fields in both these data objects. Select COMPOSITEINSTANCEID from the list under Fields to match from Lookup Data Object  and COMPOSITE_INSTANCE_ID from the list under Fields to match from this Data Object. Click Add to define the lookup criterion. More than one such lookup criterions can be added if needed be. Similarly add another lookup field i.e the STATE from the TaskDetails data object. Keep the lookup criterions based on the COMPOSITE_INSTANCE_ID and COMPONENT_INSTANCE_ID. Now the  BI_default_OnlineAccountOpening_AccountOpeningProcess will also have the values for TASKID and STATE for a given composite instance.

Configuring the Streaming Action List for Inflight Instances

Make sure everything is saved for the moment. Edit the  original report template and for the next horizontal tile too select the view type as Action List. The streaming list view will have to be now configured to show the relevant in-flight live instances. Open the report in the Edit mode. For the data object field select the same data object as used in the previous view i.e BI_default_OnlineAccountOpening_AccountOpeningProcess. Go to the Fields tab to select and arrange the following fields in their order: COMPOSITE_INSTANCE_ID, BI_NAME, COMPONENT_NAME, METRIC_cardApplicationStatus, STATE, TASKID. From the Sort tab add BI_NAME and METRIC_cardApplicationStatus under the Sorted Fields.

As stated earlier if no filters are applied to this data object list, it will end up displaying all the instances against all executed activities in the business process. In order to view non-duplicated composite instances only the ones that are latest are to be filtered out. and just those that are assigned to process users. Having such filters is easy. Click on the Filter tab and click on Add Entry. This opens the Filter Expression wizard. Add a filter for the STATE field having a value of ASSIGNED. Similarly add another filter for the LATEST field to have a value of Y. This would do the work.

The display settings of a report are configured from the menu tabs under for the Properties button. Click on the Properties button and under the General tab enter Live Instances for the View Title. By default the streaming action list shows checkboxes against all the rows in the report. Change these to radio buttons so that only an instance is selected at a time. Go to the Actions tab and select Single Select (radio buttons) options.

Once these settings are applied and saved the streaming list will show the in-flight instances along with their task identifier and the pending action. Optionally, the task creators, assignees, task expiration time, priorities etc. can also be shown in this list to have a better visibility as to which tasks are assigned to which process and which tasks needs urgent attention. Also note that the column titles displayed in the screenshot below is different from the data objects fields. This can be done by overriding their default names from the Text and Align tab under the Properties button.

Define an Action Button for the selected instances of the List

The final piece in the hook is to define an action button for a selected row in the report view. What is desired here is that when a row is selected the action button should take the users to the approval form for the selected instance. This would effectively mean launching the business process workspace view of the task form allowing it to be actioned if needed be. Opening the task approval form is easy. The URL to be used to launch the task form simply needs the HOST and PORT on which the business process workspace is running and the task id of the composite instance (which is obtained already and displayed in the list).

http://HOST:PORT/integration/worklistapp/faces/home.jspx?taskId=TASKID

The same set of steps that were discussed earlier to create an action button can be executed once again to create another action button. Open the report once again in the Edit mode, select the view and click on Change Report Properties.  Click on the Actions tab and then the New button to open a new action editor wizard. Name this action button as View Approval Page. In the next screen define the type of action as Open a URL. The URL for the action button can be configured by adding constant or dynamic strings (here referred to as Terms) which when concatenated will give the complete URL string.

Create a New Term with constant value by passing the entire URL string except the TASKID in it. Be sure to replace the value of the HOST and PORT with that of the server instance running the business process workspace user interface. Add another New Term but this time choose Value from a List View from the mapping type action list. The next screen displays the report skeleton from where select the live instances view in the report. This will show all the selected data object fields configured for the list as dropdowns under Select a column from this view column. Select TASKID and then click OK.

In the following Button Formatting screen simply select the Display the button in a view radio button under the Button Location panel. This will add the action button in the same view. For rest of the screens simply click on OK to save the report property changes.

Appreciating the End Result

Well, that’s it! After all these changes your BAM users will be able to see two streaming lists of completed/historical instances and in-flight instances. The users will also be able to view the instance audit trail by clicking on an instance in the list and if they happen to be process users then also action in-flight instances by selecting it and clicking the View Approval Page action button. This also allows you to create a seamlessly correlated view of instances across the multiple dashboards that Oracle SOA Suite 11g has to offer. It is important to note that not everyone in the organization will be able to view the task-form for the instance. Only users who are assigned to the process roles to action the task will be able to do so. This ensures that the security aspects of this implementation methodology is also covered

Downloadable Resources for the Project

• BAM Data Objects and Reporting Dashboard

You can import these data objects and reports into your BAM server by running the ICommand utility. The easiest way is to copy all the dataobjects and reports in the DataObjects.zip file to the <MIDDLEAWARE_HOME>\Oracle_SOA1\bam\bin directory and then run  icommand –cmdfile import.xml in the command line. Before running the Icommand utility make sure that the BAMIcommandConfig.xml file under the <MIDDLEAWARE_HOME>\Oracle_SOA1\bam\config directory is configured to point to the correct instance of the BAM server. A detailed explanation of working with the ICommand utility can be read here at http://docs.oracle.com/cd/E14571_01/integration.1111/e10224/bam_app_icommand.htm

•  

  Composite Project with the BPM Process

•  

  SOAP-UI Project to create Instances of the Project

Feel free to post comments and suggestions, and possible improvements and I will be glad to incorporate them here. If you have a similar requirement and are struggling to re-create a reporting dashboard similar to what is proposed, send me a message.

Hope this article helps!

.

Oracle SOA Suite 11g: Advanced Administration Topics

A few days ago, i had the pleasure to announce the release of my upcoming handbook on Oracle SOA Suite 11g Administration Handbook. It is my privilege once again to let everyone know that Chapter 10 of our book, that deals with some advanced administration topic is available as a free download from the publisher’s website. Here is the link for the chapter.

http://www.packtpub.com/sites/default/files/downloads/6082EN_Chapter10_Advanced_Administration_Topics.pdf?utm_source=packtpub&utm_medium=free&utm_campaign=pdf

In addition to this chapter, the sample code for the book, containing a handful of administration scripts, can also be downloaded from

http://www.packtpub.com/support/10082

The entire table of content of the book can be see here

http://www.packtpub.com/oracle-soa-suite-11g-administrators-handbook/book

I am thrilled by the number of emails and accolades I have received so far with respect to the contents of the book. I am sure this would be useful to both Oracle SOA Suite 11g Administrators and Developers alike.

Oracle SOA Suite 11g Administrator’s Handbook

Your blogger, for the past few months has been involved in what I am sure many will agree is a nightmarish activity.  As a result the blog posts have dried up a little as blogging involves both time and novelty and I have been a little short on the first lately. What has consumed me all the while? Well it was co-authoring a book on Oracle SOA Suite 11g Administration and ensuring that it met our vision to provide readers with a definitive administration reference handbook. The book was announced a month ago and is finally out and available.

My fellow co-author Ahmed Aboulnaga, has been quick enough to mention this on this widely popular blog and you can read about the book here at: http://blog.ipnweb.com/2012/08/oracle-soa-suite-11g-administrators.html

Ahmed, in his blog post, has already listed down, in details, as what is this book about, who is it for, and the contents of the book. I will thereby refrain myself from repeating the same things again here. However, I would though like to share a few insights that went into making the books and why do we think this book would be useful for the readers.

Oracle SOA Suite 11g is the backbone of messaging and application integration in a service oriented/event driven architecture. Due to its sheer size, understanding the underlying components, services, configuration and their relations, can be a daunting task. But it doesn’t have to be! Faced with the many challenges and common pitfalls involved in real-world Oracle SOA Suite implementations, we realized an opportunity to provide a comprehensive yet practical solutions guide to both re-actively and pro-actively manage the infrastructure.  We have ensured a lot of our real-world
experiences of working with Oracle SOA Suite 11g infrastructure are captured in the book and that we deliver a comprehensive guide filled with grounded solutions, best practice(s), and pragmatic instructions.

We begin by providing an introduction of SOA and quickly move on to provide as much concise and useful information related to management of SOA composite applications. We feel it is important to know how to manage composite applications, their deployments and lifecycles to be successful in administering them.

First and foremost, this book is an attempt to equip the reader with everything they need to know about all the core administrative and management activities around Oracle SOA Suite 11g. Our attempt throughout this book has been to give readers administration tips straight from the trench, i.e from our real time experiences. Among the many things covered in this book are automation techniques to manage the infrastructure, troubleshoot root causes of error, devising strategies of reuse with respect to code, configuration and customizations, engage in performance tuning of each components individually or the infrastructure as a whole apart from routine administration activities. The book has a wealth of information that can be applied to real world scenarios or help you understand the best ways to tackle core administration tasks in Oracle SOA Suite 11g. It has been structured to allow readers to open up any chapter of interest, with earlier chapters beginning with more basic administrative concepts while moving to more advanced topics towards the latter half.  Right from Chapter 1, we have made the best effort to ensure that the reader is taken through a step-by-step, tutorial based approach offering a unique insight into the what, why and how of all administrative related activities involved with Oracle SOA Suite 11g. Our aim throughout was to put forward a definitive guide to real world administration of Oracle SOA Suite 11g, that provides a monitoring, troubleshooting and tuning methodology, explains what and how to backup and restore your environment, configure the infrastructure to implement a policy based security framework, offer resuable scripts for common administration tasks such as deployments,tuning and migration, and delves into advanced topics such as silent installs, cloning, backup and recovery and provisioning highly available infrastructure.

Instructions and guidelines provided in the book offer a lot of handy tips and best practices, all based on real-world problems and solutions that Oracle SOA Suite 11g Administrators can benefit with and significantly increase their productivity. We certainly feel that you will feel armed with all the “to-do’s” and “know-how’s” of basic and advanced administration with this book. The skills and insights learnt from each of the chapters in this book can inadvertently be  applied to real world administration.

Part of our writing style in this book draws heavily on the philosophy of reuse and as such we provide an ample of executable SQL queries and WLST scripts that administrators can reuse and extend to perform most of the administration tasks such as monitoring instances, processing times, instance states and perform automatic deployments, tuning, migration, and installation. These scripts are spread over each of the chapters in the book and can also be downloaded from http://www.packtpub.com/oracle-soa-suite-11g-administrators-handbook/book

I would finally try to make a brave and best attempt to summarize in brief the key learning’s from the book:

• Overview of the Consoles and Key Administration Areas.
• Monitoring Instances, Messages, and Composites
• Configuring Audit Levels and means of end-to-end monitoring through the use of extended logging
• Administering and configuring components within Oracle SOA Suite 11g such as the BPEL, Mediator, BAM, EDN engines, Services and References.
• Monitor and Performance Tuning service engines, WebLogic server, threads and timeouts, files systems, and composite applications.
• Interacting with and configuring the infrastructure MBeans using both Oracle Enterprise Manager Fusion Middleware Control and WLST based scripts.
• Troubleshooting approaches as how to identify faults and exception through extended logging and thread dumps and find solutions to common start-up problems and deployment issues.
• Secure the components deployed to the infrastructure by leveraging Oracle Web Services Manager along with the supporting infrastructure configurations.
• Later chapters deal with managing the Metadata Services Repository, dehydration store, backup and recovery.
• Use case depicting entire steps involved in upgrading from one patchset to another and migrating from SOA Suite 10g to 11g.
• Advanced topics such as silent installations, cloning, and high availability installations.

Finally, I would reiterate the statement made by my co-author once again that this book is not a re-hash of Oracle’s documentation on Oracle SOA Suite 11g. It offers insights into SOA Suite administration that you will not find elsewhere. If you are an Oracle SOA Suite Administrator or aspiring to be one, then this book is your bible. It gives you everything you need to about all your tasks and lets you apply what you learn in your everyday life right from chapter.

The book is available at the following websites:

Paperback and eBook versions:
http://www.packtpub.com/oracle-soa-suite-11g-administrators-handbook/book

Kindle version:
http://www.amazon.com/Oracle-Suite-Administrators-Handbook-ebook/dp/B0092JYZXI/ref=tmm_kin_title_0?qid=1346074747

Any feedback, comments and inquiries are most welcome!

Arun Pareek

 

Custom XPath Functions in OSB 11g to Lookup Shared DVMs

The Problem Statement(s)

I was recently working on a big OSB implementation for one of our clients using the Oracle SOA Suite 11g platform. Interestingly we had to use both OSB and components in the SOA Suite 11g platform to cater to different needs integration and business scenarios. However one common thing that we needed to use across both the components were DVM’s and XREF’s. It would be naive to highlight this as a problem as this is well known and well blogged about. There has been numerous blog posts showing workarounds to implement DVM alike functionality in OSB. However none of them seemed to work for our implementation. To highlight a few reasons, here were our guidelines and the challenges thereof:

• All DVM’s were deployed as a shared and reusable project into the MDS. Needless to explain why we did this. Once we deploy DVM’s as a shared project to the MDS they become available from the SOA Composer and can be edited/changed at runtime.
• These DVM’s were to be used both across OSB and BPEL/Mediator to lookup values. Now obviously there is no direct way to access DVM’s deployed as a shared project or within a composite from OSB, so we relied on creating a composite, exposed as a service that will be invoked by the OSB, have a mediator look up the DVM and return the response.
• This workaround worked for us but with the baggage of hundreds of calls being made between OSB and SOA Suite service just for DVM lookup retrievals.

Well I am sure everyone will have to say that DVM like functionality will be introduced in the future releases of OSB. But what about till then? Also once Oracle releases a new version of OSB, say 12c, do we expect every 11g implementation in the world to migrate over to it.  The question that lead me to ponder over and thereby resulting in this article was “Is it that difficult to reuse the existing DVM XPath functionalities (and the DVM’s that are deployed to the SOA infrastructure) in OSB?”. Turned out it is difficult but with a little effort, I was able to put the pieces together and will explain in this blog, in a step by step way, how was it all done.

Prerequisite(s)

• Oracle SOA Suite 11gR1 (PS2 and higher)
• Oracle Service Bus 11gR1(PS2 and higher)

Creating and Deploying the Shared DVM Project

Let me start by creating a very basic composite containing just a DVM (CountryCode.dvm). This is good enough to test all the three different DVM functions available in Oracle SOA Suite 11g. The DVM just store a couple of records of Country names, their Country and Currency codes. The SharedDVMProject is then deployed to the SOA Infrastructure so that the DVM is available in the MDS.

Once the resource gets deployed to the SOA infrastructure the easiest way to locate it externally is by accessing it over http. The deployed DVM should be available at http://<host>:<port>/soa-infra/services/<partition>/<compositeName>/CountryCode.dvm. Replace these values of host, port, partition and the composite name with values belonging to your environment and open the address in a browser to see the DVM xml.

Oracle SOA Suite 11g supports primarily three ways to look up DVMs. For a full read on working with DVMs refer to the link http://docs.oracle.com/cd/E14571_01/integration.1111/e10224/med_dvm.htm.

The three available XPath DVM functions are

• dvm:lookupValue(dvmMetadataURI as string, sourceColumnName as string, sourceValue as string, targetColumnName as string, defaultValue as string) as string

The above function is used to return a looked up value of the string by looking up the value for the target column in a domain value map, where the source column contains the given source value. If the target value is not found the default value is returned.

• dvm:lookupValue(dvmMetadataURI as string, sourceColumnName as string, sourceValue as string, targetColumnName as string, defaultValue as string, (qualifierSourceColumn as string, qualifierSourceValue as string)*)) as string

This function is also used to retrieve a looked up value of the string by looking up the value for the target column in a domain value map, where the source column contains the given source value. If however you dvm contains qualifiers columns then you can pass multiple qualifier columns and their values additionally to the DVM.

• dvm:lookupValue1M(dvmMetadataURI as string, sourceColumnName as string, sourceValue as string, (targetColumnName as string)? ) as nodeset

This one to many dvm function returns an XML document fragment containing values for multiple target columns of a domain value map, where the value for the source column is equal to the source value. You can provide all the target columns for which the values needs to be retrieved as comma separated string names.

The arguments as explained in the oracle documentation site for these various functions are:

• dvmMetadataURI – The domain value map URI. Here in our example you could simply pass CountryCode.dvm
• sourceColumnName – The source column name.
• sourceValue – The source value (an XPath expression bound to the source document of the XSLT transformation).
• targetColumnName – The target column name.
• defaultValue – If the value is not found, then the default value is returned.
• qualifierSourceColumn: The name of the qualifier column.
• qualifierSourceValue: The value of the qualifier.
• targetColumnName – The name of the target columns. At least one column name should be specified. The question mark symbol (?) indicates that you can specify multiple target column names.

Importing the Custom DVM XPath Library into OSB

Having said, this, we would probably like to see a support for all these three dvm functions if we are building a custom XPath library for them in OSB. Well the Custom DVM XPath library that I have created, and that you can use, provide an implementation of all the three functions. However the usage semantics is a little different and I shall explain how each of the function can be used. Before proceeding further, if you want to know how custom XPath libraries can be added to both OSB designtime and runtime refer to these blogs that contain some useful demonstration about it.

https://beatechnologies.wordpress.com/2010/09/08/custom-xpath-in-osb-11g/

http://docs.oracle.com/cd/E14571_01/doc.1111/e15866/custom_xpath.htm

Once you know the basics of how custom XPath functions can be made available to OSB, download the osb-dvm-xpath.jar available from the link here and copy it in <MW_HOME>\Oracle_OSB1\config\xpath-functions folder.

You would also need a .xml (here custom-osb-dvm-xpath.xml) file, as shown below to describe the usage of these functions. The xml file basically contains how would you like to name your function and also which class in the package has a method to implement the corresponding function.

<xpf:xpathFunctions xmlns:xpf="http://www.bea.com/wli/sb/xpath/config">
<xpf:category id="DVM XPath Functions">
<xpf:function>
<xpf:name>lookupDVM</xpf:name>
<xpf:comment>Lookup Value DVM Function based on dvmLocation,source column, source value, target column and target value</xpf:comment>
<xpf:namespaceURI>http://www.com.soatechnologies.blog/osb/custom/xpath/dvm-xpath-functions</xpf:namespaceURI>
<xpf:className>blog.soatechnologies.osb.dvm.DVMXpath</xpf:className>
<xpf:method>java.lang.String lookupValue(java.lang.String, java.lang.String, java.lang.String, java.lang.String, java.lang.String)</xpf:method>
<xpf:isDeterministic>false</xpf:isDeterministic>
<xpf:scope>Pipeline</xpf:scope>
<xpf:scope>SplitJoin</xpf:scope>
</xpf:function>
<xpf:function>
<xpf:name>lookupDVMWQ</xpf:name>
<xpf:comment>Lookup Value DVM Function based on dvmLocation, source column, source value, target column, target value and qualifiers</xpf:comment>
<xpf:namespaceURI>http://www.com.soatechnologies.blog/osb/custom/xpath/dvm-xpath-functions</xpf:namespaceURI>
<xpf:className>blog.soatechnologies.osb.dvm.DVMXpath</xpf:className>
<xpf:method>java.lang.String lookupValueWQ(java.lang.String, java.lang.String, java.lang.String, java.lang.String, java.lang.String,org.apache.xmlbeans.XmlObject)</xpf:method>
<xpf:isDeterministic>false</xpf:isDeterministic>
<xpf:scope>Pipeline</xpf:scope>
<xpf:scope>SplitJoin</xpf:scope>
</xpf:function>
<xpf:function>
<xpf:name>lookupDVM1M</xpf:name>
<xpf:comment>Lookup Value DVM Function based on dvmLocation, source column, source value and multiple target column qualifiers</xpf:comment>
<xpf:namespaceURI>http://www.com.soatechnologies.blog/osb/custom/xpath/dvm-xpath-functions</xpf:namespaceURI>
<xpf:className>blog.soatechnologies.osb.dvm.DVMXpath</xpf:className>
<xpf:method>org.apache.xmlbeans.XmlObject lookupValue1M(java.lang.String, java.lang.String, java.lang.String,org.apache.xmlbeans.XmlObject)</xpf:method>
<xpf:isDeterministic>false</xpf:isDeterministic>
<xpf:scope>Pipeline</xpf:scope>
<xpf:scope>SplitJoin</xpf:scope>
</xpf:function>
</xpf:category>
</xpf:xpathFunctions>

As you can see that the the XML file has three functions, namely lookupDVM, lookupDVWQ, lookupDVM1M that mimic the ones available in Oracle SOA Suite 11g, albeit with a little difference.

The three available XPath DVM functions are

• dvmG:lookupValue(dvmLoc as string, sourceColumnName as string, sourceValue as string, targetColumnName as string, defaultValue as string) as string

The above function is exactly the same as the one available in Oracle SOA Suite library. It returns the looked up value for the target column.

• dvmG:lookupValueWQ(dvmLoc as string, sourceColumnName as string, sourceValue as string, targetColumnName as string, defaultValue as string, qualifierArray element(*)) as string

This function enables a dvm lookup by providing the necessary qualifiers, only in this case rather than passing multiple qualifier columns and their values as sting arguments, the function accepts a Document containing a list of qualifiers.

• dvmG:lookupValue1M(dvmMetadataURI as string, sourceColumnName as string, sourceValue as string, targetColumnArray as element(*) ) as element(*)*This one to many dvm function returns an XML document fragment containing multiple matched values for all the queried columns for a particular source column/value pair. All the target column names should be passed as an element array instead of repeating string arguments instead.

The arguments as explained in the oracle documentation site for these various functions are:

• dvmLocation– This is similar to the domain value map URI. However since the DVM will not be present in the OSB runtime but instead be deployed to the SOA infrastructure the URI has to be http://<host>:<port>/soa-infra/services/<partition>/<compositeName>/CountryCode.dvm instead of just CountryCode.dvm, i.e the DVM name.
• sourceColumnName – The source column name.
• sourceValue – The source value (an XPath expression bound to the source document of the XSLT transformation).
• targetColumnName – The target column name.
• defaultValue – If the value is not found, then the default value is returned.
• qualifierArray : XML Document containing repeating arrays of qualifier name and column similar to the fragment shown here: <qualifiers><qualifierName>CurrencyCode</qualifierName><qualifierValue>GBR</qualifierValue></qualifiers>. However note that you may use any names for the XML elements but the structure needs to similar.
• targetColumnArray – An XML Document containing a list of repeating of the target columns. For example, the input should be somewhat like <targetColumns><column1>CountryCode</column1><column2>CurrencyCode</column2></targetColumns>. Here also the name of the elements being used can be anything but the structure of the Element array has to be quite like above.

Once the custom-osb-dvm-xpath.xml and osb-dvm-xpath.jar are placed in the <MW_HOME>\Oracle_OSB1\config\xpath-functions start/restart your Eclipse IDE and the OSB server to see the Custom DVM XPath library made available to the OSB functions. The following image also shows a Xquery file dvmLookup.xq that uses the custom library function dvmaG:lookupDVM by making all the arguments required for the function as externally passed values from users.

Once done, execute this on the server by right clicking on the file and Run As –> Run on  Server. This brings up the OSB Test console for XQuery Resource Testing. Pass some valid values to look up a particular DVM target column value as shown below to see the response.

Similarly create two more XQueries to test the other two functions. Or alternatively, import this soa.osb.dvm.project.jar that has all the XQueries created for you to test. Look for how the qualifiers are passed in the image below while testing the dvmLookupWQ function. The document map is convenient to create as you can pass as many qualifiers name value pairs under the root document.

The lookupDVM1M is a bit different from the functions above as it returns document fragment containing multiple target columns for which the values needs to be retrieved all at once. This is useful in cases where you need to retrieve multiple target column values at once. The only trick part is that the target columns should be placed inside repeating elements inside a root level document while being passed as arguments.

Now all was good until now as I could use already deployed DVMs from the OSB itself by just a few XPath functions that provide an exact similar functionality as the ones in Oracle SOA Suite. All was good until I further realized that sending a complete URL for the MDS is a pain. For instance when we use the DVM functions in SOA we just send the DVM name as in CountryCode.dvm but to use the corresponding functions in OSB, i had to use :/soa-infra/services///CountryCode.dvm”>http://<host>:<port>/soa-infra/services/<partition>/<compositeName>/CountryCode.dvm instead. This turned out to be another problem as the environments keep on changing, and so do the partitions that these DVMs are deployed to. Also there were a few hundreds of DVMs that were used and hence using the full URL to access the DVM was not accepted and ruled out.

In the next step, I refined the Custom XPath functions to take just the DVM name instead of the entire URL. But the question is how would the OSB runtime locate the DVM’s that are deployed somewhere in the SOA MDS. Well, turns out all I needed to do is write another utility to look for and extract the DVM from the MDS based on the DVM name.

The Custom DVM classes just needed the MDS Database connection information to connect at runtime and look for the DVM metadata to retrieve it. I simply created a db.properties file, something like what is shown below, that has some name value pairs to describe the database connection details. This file can be edited to replace the values with the ones corresponding to your environment and copied to the <MW_HOME>\user_projects\domains\<Domain_Home> folder of your OSB domain.

username=<MDS User Name>
password=<MDS Password>
dbURL=jdbc:oracle:thin:@<DbHost>:<DbPort>:<DbServiceName>
partitionName=<Partition Name>
connName=<Any arbitrary Name for the MDS Connection>

So next time when you test any of the above DVM function, you can either specify the complete URL (or the HTTP address where the DVM is located), or just the DVM name only.

Feel free to post comments and suggestions, and possible improvements and I will be glad to incorporate them here. If at all you have a similar requirement or are facing problems similar to what I faced and would like to get the binaries of the jar, send me a message. Hope this helps!

Once again, the files used in this example can be downloaded from the links below

osb-dvm-xpath.jar

soa.osb.dvm.project.jar

.

Understanding Event Based Gateways in Oracle BPM Suite 11g

In my previous blog post I had covered an example and use case to use Complex Gateways in Oracle BPM Suite 11g. Complex Gateways allows you to implement a special type of dead path elimination modelling and may be very useful in scenarios where it is required to synchronize N paths from M incoming transitions. The full post can be accessed at the link here : https://beatechnologies.wordpress.com/2012/03/14/using-complex-gateways-in-oracle-bpm-11g/

In this post, I will try to provide a pattern, example, use case and practical demonstration to use an Event Based Gateway, which is yet another important modelling concept in BPMN based process. Apart from the usual stuff I would also share a few interesting observations on Interrupting events that may be useful among other things.

Knowhows of an Event Based Gateway

Event based gateways are a special case of deferred choice pattern where the process chooses a specific sequence and branch depending on the interaction with an external environment thereby allowing for true dynamicity. This is different from the exclusive gateway in the sense that it allows a branching point where alternative paths are determined based on intermediate events coming from external systems. The message flow waits at the event based gateway waiting for a token to arrive from any of its branches. The token can be a receipt of an event message, a time cycle or a receive task. The first event triggers one of the alternative to the exclusion of any other path from the gateway. Essentially, the Event “pulls” the Token from the Gateway, where it will continue down the sequence flow that leaves the Event.

Introducing the Use Case

The use case explained here in this blog is that of a simple banking process to automate online opening of Bank Accounts. The process initiates with an application request being received from an Online Portal and verifying if a customer has already made an application earlier. If a similar request is already in process it is rejected and an appropriate notification is sent to the customer. Otherwise the process generates a provisional Account/Application Id and initiates a process to mail the account starter and contract kit to the address received from the customer. The customer is additionally sent an email with the necessary details and asking him to send back the proof of identity related documents and a signed copy of the contract within 7 days. This is where the process waits to receive documents from customer or remind him again after 7 days. The event based gateway branches into two sequence; 1) an intermediate event that receives an acknowledgement of customer documents being received and 2) a timer based event that waits for 7 days, sends a reminder email to the customer and loops back into the gateway.

Notice the illustration of the BPMN process for AccountOpeningProcess above has two event based gateways.

1. Process Online Application
2. Receive Supporting Documents

The Process Online Application gateway is a gateway of instantiating type meaning that it doesn’t need any incoming sequence flow. This kind of a gateway is useful if there are multiple types of messages or events that can start a business process. To demonstrate this use case follow the Event subprocess for cancelling in-flight applications. Whenever a cancellation event is received intermittently within a process an exclusive gateway decides whether a CustomerExperienceExecutive should make a personal call and deter the cancellation. If a resolution is reached i.e by means of changing the account type, account charges and privileges, the main process is re-initiated. Thus the process ends up having multiple start events, one that is for new applicants and one that is for re-applicants for whom the CheckCustomerSummary step may be skipped.

On the other hand the Receive Supporting Documents gateway is non-instantiating type and hence has both incoming and multiple mutually exclusive outgoing sequence flows. The event based gateway after the task indicates that the process actually waits for two different events that could happen next: Either the acknowledgement is received from the process handling document acknowledgements from the customer, as indicated with the Supporting Documents Received message event, or there is no acknowledgement for 7 days after which a reminder is sent to the customer.

It is important to know that Event based gateways with one outgoing Event/Receive Task and one timer event can be modelled in a different way as well, thereby completing the need to have a gateway at all. The timer boundary event can be used on the Receive task as shown under. But anyways the idea here is to show how event based gateways work in real life use cases.

 

Prerequisite(s)

The modelling workspace and the version of Oracle BPM Suite 11g used in this demonstration is 11gR1PS5 i.e 11.1.1.6.

Modelling the Process

Follow the BPMN model explained in the use case and recreate it in JDeveloper BPM Studio. Else the modelled process can be downloaded from the link here.

For the sake of reusability I created a reusable process to send email notifications that can be invoked from multiple places in the main process using a Call Activity.

A Call Activity identifies a point in the Process where a reusable global Process is used. The Call Activity acts as a wrapper for the invocation of the reusable process by transferring the control to the called reusable process. A Reusable Process on the other hand is a reusable, atomic process definition that can be called from within any Main Processes by a Call Activity. The reusable process in this example determines the content and subject of the customer notification by using a rule activity and sends an email notification accordingly.

Implementing the Process

Open the downloaded application workspace for the project or if you have created a one of your one, then double click on the Process Online Application event based gateway and in the Implementation tab check the Instantiate checkbox.

The first thing required is to add all the Request and Response types from the schemas in the xsd folder as business objects in the business catalog of the process. The two message events sequences for new applications and reapplications are based NewAccountRequest type and the Send Application Status end event is based on the AccountApprovalResponse type.

Take a close look at the three events intermediate events in the process:

1. Supporting Documents Received
2. Cancel Application
3. Reapply

All of these intermediate events receive some form of communication in the middle of the process and hence has to correlate with the actual instance of the main process. If a correlation is not defined then two things can happen. Either a mid-process receive will create its own instance or there will be an error in the process due to conflicting receive. Hence the message start events has to initiate a correlation and these intermediate events need to use them.

For the purpose of this example, the taxNumber of the applicant is used as the correlation key as it is assumed that this should be unique.

And similarly in the intermediate message receive events, the same correlation key needs to be used to relate the incoming request to existing instances. So when an external system sends the acknowledgement when supporting documents from the customer has been received the message is correlated back to the original instance.

The rest of the process is easy to implement. Activities that cannot be implemented may marked as draft and the exclusive gateways can have conditions that favour the happy path of the process. All the Call Activities invoke a reusable process that run a business rule to populate the Email subject and body and sends an email. Creating reusable processes is an attractive pattern and should be widely adopted as this allows to keep fragments of common functionalities globally that are accessible for multiple processes.  A sub-process or an event sub-process on the other hand may contain functionality that is common in a particular process and may be accessed from multiple process points.

The implemented process and the JDeveloper workspace can downloaded from here in case you want to spare yourself from the work and go straight to test and see results.

Testing the Process and Seeing Results

New Application Request

Deploy the implemented process to a BPM domain. The process has an event gateway of instantiation type that waits for either a new application request or a reapplication event message to arrive. A business process that has a message start event like this, has a service endpoint and can be invoked as a standard soap based service. Use any SOAP testing tool (SOAPUI for instance) to invoke the business process operation that accepts new account opening application request.

The Process Online Application event gateway initiates the sequence that has the New Application message event and the instance stalls at the Receive Supporting Documents event gateway. The instance status will be Running and the token will stay here until an acknowledgement is received from a support process or the waiting time exceeds seven days.

The process Flow trail gives another diagrammatic view of the sequence of activities executed for this instance.

Now for the sake of testing the intermediate event based gateway, the Supporting Documents Received message is sent to the process with the same taxNumber as the original request and the event will correlate it with the running instance.

Executing the above request will result in another instance ID being generated in the BPMN service engine but this instance correlates with the already in flight one. Since the documents are successfully received within the given time frame the token moves from the event based gateway to the Supporting Documents Received message event and consequently completing the instance.

Check out the flow view to see a graphical trail to see how the instance executes forward from the event based gateway and completes.

Cancel Application Leading to Reapplication

The above example show how an intermediate event based gateway behaves. Now let me demonstrate how an event based gateway of instantiation type works. For this example, I will again invoke the process with a new application request, but while the process is waiting to receive documents from the applicant, a Cancel Application event will be fired. This is picked up by the Cancellation Event Subprocess and correlated to the running instance. The subprocess evaluates the reason for cancellation and if required creates a task for a CustomerExperienceExecutive role to personally call the customer and seek a resolution. If a settlement is reached i.e by altering the account charges, privileges or type etc., the instance is sent back for reapplication. The Process Online Application at the beginning of the main process has a sequence flow that waits for this kind of event that picks up the message and continues the process flow, only this time there is no check performed to check client history.  The first step is to create a new request once more and then invoke the cancellationRequest operation.

In the Composite dashboard, there will be two instance of the process now, one that has completed (original) and one that is running (reapplication).

The first instance is marked as Completed as the Cancel Application Event is an intermediate event of Interrupting Type and hence when it reaches an End activity the actual instance itself is marked as completed.

The audit flow shown below gives a better picture of what happens to the instance in this case.

Now inspect the other instance that is marked as Running. This spawns from the Reapply intermediate event and this time the gateway branches to the sequence that has the Reapplication Event Message type.

The flow trace exactly reveals this behaviour where a new instance is created, only this time with a different message types and executes upto the next gateway.

I made some following interesting observations while implementing and verifying the test results for this process that I would like to share.

1. If the Cancel Application throw intermediate event in the Event Subprocess is of non-interrupting type and the end event is of Terminate type, the process behaviour should ideally be the same. But it is not. While a new instance spawns up correctly but the original instance is marked as faulted.
2. Since there is a process service call being made from the Reapply throw event that is caught by the Reapplication message catch event, it is not necessary to create a correlation between these events. If a correlation is not defined, the engine should mark the original instance as completed (as it flows to the End event) and create a new independent instance from the Process Online Application gateway. But there is a little deviation here. In this case both the instance states are marked as Completed in the dashboard view but when drilled down to the audit trail, the process instance is shown to be running and waiting at the Receive Supporting Documents gateway.

This article comes to a conclusion here and hopefully it was able to provide an insight into how Event based gateways can be used in Oracle BPM Suite 11g and also some thoughts on intermediate events. Event based gateways are extremely handy while designing loosely coupled business processes using a deferred choice pattern.

The JDeveloper workspace and sample requests to execute this Oracle JDeveloper/SOA Suite 11.1.1.6 is available to download from here.

Feel free to provide comments, suggestions and feedback for this post and I will be happy to do my best.