Saturday, September 4, 2010

Workflow Policies vs Workflow Processes

Building a solution with Siebel Workflow often involves the use of both Workflow Policies and Workflow Processes. Siebel Workflow, taken together, is a complete application for automating server processes defined using declarative relationships between logical objects. As a Siebel Developer, you need to understand the difference between a Workflow Policy and a Workflow Process.

A Workflow Process is a program that runs on the Siebel server. It is defined through a graphical interface as a set of steps. When the process runs, a single record is processed. The workflow process steps are performed as a series of data operations.

A Workflow Policy is a specific event that occurs on the Siebel database. Based on a database trigger, it can include many complex criteria, but it ultimately evaluates to a true/false condition to determine whether to execute a program or not. Commonly, a Workflow Policy will execute a Workflow Process.

Do not be confused between the Business Object that is part of the Workflow Process definition and the Workflow Policy Object that is part of the Workflow Policy definition. A Workflow Process runs on the business layer of the Siebel object model. The Business Object that helps define a Workflow Process is the same Business Object that governs the logical data entity relationships between Business Components in Siebel screens and views.

A Workflow Policy Object is also configured in Siebel Tools, and it also represents a logical data entity, but it seems closer to the data layer of the Siebel object model. Workflow Policy Objects, Components, Columns, and Component Columns are a objects that do not contain or enforce any business rules. They are essentially columns and tables, and the relationships between them.

By first understanding the basic differences between these two core components of Siebel Workflow, a Siebel Developer can begin to grasp the basics of the powerful business process automation application known as Siebel Workflow.

Monday, April 5, 2010

Abstracting Database Passwords in Batch Scripts

Even when a Siebel implementation does not need to be SOX compliant, it is still important to develop and maintain processes to reduce errors and fraud. Separation of duties (SoD) is an important security principle in any enterprise application environment. For example, it is often best to prevent Siebel Developers from having administrative access, and to prevent Siebel Administrators from changing code.

One potential vulnerability is that command-line server manager connections require a username and password that authenticate against the Siebel database. People with this information can use a third-party tool to access and manipulate the Siebel database. In a production environment, administrators need these passwords, but they should be restricted as much as possible, especially from developers.

Scripts invoking the Siebel Server Manager command-line interface can be a powerful tool for automating server tasks, but connecting to the command-line interface on a Windows server requires the following syntax:
srvrmgr /g gateway1 /e enterprise1 /s server1 /u sadmin /p sadmin
In the above command the /u and /p arguments require a valid username and password using database authentication. A batch script containing this information challenges the SoD principle. Either an administrator manipulates the script to insert the password, or a developer does. Either way, the roles become blurred.

The solution to this problem is to isolate passwords and other environment-specific information from the script itself.

Consider the following excerpt from a Windows shell script:
call E:\secure\envvariables.cmd

E:\sba80\siebsrvr\BIN\srvrmgr /g %gateway_server% /e %enterprise_server% /s %siebel_server% /u %eimuserid% /p %eimpassword%
In the envvariables.cmd file, the following:
@set gateway_server=PRODGTWY
@set enterprise_server=Siebentprod
@set siebel_server=Siebprodbat1
@set eimuserid=EIMIMPORT
@set eimpassword=SecurePwd
It doesn't matter how much complex logic is added to the shell script containing the srvrmgr command, user names and passwords are segregated from the logic in a file that can only be modified by the system administrator. Moreover, environment information is also segregated, so the script can be migrated through UAT and Production without modification.

Friday, March 12, 2010

Interview Question #2 - What is a Siebel Operation Step?

This interview question uses a technical term to test a Siebel Developer's understanding of a topic. "Siebel Operation" can be almost anything to someone who does not have a basic familiarity with Siebel Workflow, but it is an everyday term for any Workflow Developer.

Q: Please explain what a Siebel Operation is, and how it is used.

A: At minimum, the candidate should know that a Siebel Operation is a type of Workflow Process Step. If the candidate does not volunteer this information without additional prompting, he or she is not a Workflow Developer.

Candidates should know that a Siebel Operation can be used to Insert or Update records as part of a Workflow Process. A candidate should know the difference between a Workflow Process and a Workflow Policy or Workflow Policy Program. Siebel Operation is a term that is only used in connection with Workflow Processes.

In addition to Insert and Update, recent versions of Siebel have other types of operations. Most Siebel Workflow Developers know that a Query operation is now available. Since Siebel 8.0, there are Upsert and looping operations: PrevRecord, NextRecord, and QueryBiDirectional. In my experience, knowledge of these operations is less common; it can be difficult to find a developer who can explain how to build a loop in a Workflow Process.

Workflow Process Steps operate on the business layer of Siebel, as opposed to the database layer. A Business Component that is associated with the Workflow Process's Business Object is required for any Siebel Operation. Workflow Developers should know these things, although a little prompting may be required.

A good Workflow Developer should also know about the Siebel Operation Object Id process property, which is updated after an Insert, Update, or Upsert operation. If one record is inserted or updated, this process property will contain the row id of the affected record. If more than one record, the property will contain an asterisk: '*'. If no records are affected, the property will not contain a value.

Sunday, February 28, 2010

VBC Compatibility Mode

The Query method of a VBC Business Service in Siebel versions later than 7.5 has an Inputs property set whose structure can be difficult to navigate. Take a look at the XML representation from Siebel Bookshelf:

<siebel-xmlext-query-req>
  <buscomp id="1">Contact</buscomp>  
  <remote-source>http://throth/servlet/VBCContacts</remote-source>  
  <max-rows>6</max-rows>  
  <search-string>=([Phone] IS NOT NULL) AND ([AccountId] = "1-6")</search-string>  
  <search-spec>  
    <node node-type="Binary Operator">AND     
      <node node-type="Unary Operator">IS NOT NULL       
        <node node-type="Identifier">Phone</node>         
      </node>        
      <node node-type="Binary Operator">=         
        <node node-type="Identifier">AccountId</node>         
        <node value-type="TEXT" node-type="Constant">1-6</node>      
      </node>    
    </node>  
  </search-spec>  
  <sort-spec>  
    <sort field="Location">ASCENDING</sort>     
    <sort field="Name">DESCENDING</sort>   
  </sort-spec>
</Siebel-xmlext-query-req>
I've found that the search-string node of the property set isn't particularly useful unless your back-end data source has a column structure that matches your VBC. In the example above, you can quickly see how difficulty it could be for a Siebel developer to write a script to parse a property set containing a search-spec node with any complexity.

In Siebel versions prior to 7.5, the Inputs property set was much simpler. Below, see an eScript snippet that unloads a search specification in the query method of a VBC Business Service using the older format of input:
var child = Inputs.GetChild(0);
var sPolicyNumber = child.GetProperty("Policy Number");
var sLastName = child.GetProperty("Date of Birth");
var sFirstName = child.GetProperty("First Name");
Where query specifications are entered into a form applet as field, the old format lets you easily retrieve input values and manipulate them in eScript variables.

For Query input in the earlier, simpler format, add the following Business Component User Property to your VBC:
Name: VBC Compatibility Mode
Value: Siebel 7.0.4

Thursday, February 18, 2010

Interview Question #1 - What is a Link?

I've decided to add a new feature to this blog. With this post, I am introducing a series of interview questions that Siebel developers and development leads should consider when preparing for technical interviews. I've interviewed many developers, and I've been interviewed quite a few times as well, and I have a pretty good idea of what makes a good technical interview question.

When I interview someone, my questions are designed to discover what a candidate knows, not bolster my ego by proving that I know something the candidate doesn't. I focus on the fundamentals of Siebel configuration, allowing the candidate to demonstrate the depth of his or her knowledge.

Q: Please describe the Siebel configuration object called a "Link".

A: The candidate should be able to provide at least two of the following, but should not contradict any of them:
  • A Link defines the relationship between Business Components.
  • Links are used to define a Business Object; the relationships between the primary Business Component and other (child) Business Components in the Business Object are Links.
  • A Link is not the same thing as a Multi-Value Link or a Multi-Value Group, but the definition of a Multi-Value Link does include a Link.
  • A one-to-many Link makes a master-detail View possible.
  • Links are defined on the Business Object layer, using Business Component Fields rather than Table Columns, although many-to-many links use Table and Column names to define the intersection table.
  • The "Source" Field is on the Parent Business Component, while the "Destination" Field is on the Child Business Component.
  • A Link can have a Search Specification.

It's ok to prompt the candidate with leading questions to develop a better understanding of the depth of his or her knowledge, asking open-ended questions wherever possible.