How does a conservative government organization fast track a paradigm shift? One method that has worked well is to keep the change process so simple that the lure of it is irresistible. This is precisely how it happened at the New York State and Local Retirement System (NYSLRS). The paradigm shift is one towards Business Rule Management (BRM). The change process is the User-defined Function Approach (UDFA). The measurement of success was provided by Barbara von Halle, founder of Knowledge Partners, Inc (KPI).
How does a conservative government organization fast track a paradigm shift? One method that has worked well is to keep the change process so simple that the lure of it is irresistible. This is precisely how it happened at the New York State and Local Retirement System (NYSLRS). The paradigm shift is one towards Business Rule Management (BRM). The change process is the User-defined Function Approach (UDFA). The measurement of success was provided by Barbara von Halle, founder of Knowledge Partners, Inc (KPI). Based on KPI’s preliminary evaluation, NYSLRS was comfortably at Rules Maturity Model (KPI RMM™) Level 2 (For details on KPI RMM™ levels 0, 1, and 2, see BPMStrategies Magazine, Premier Issue, April 2005). The meaning was clear, NYSLRS had progressed in less than one year’s time from KPI RMM™ Level 0 to KPI RMM™ Level 2. That is to say, NYSLRS took a major leap from not thinking of business logic in terms of business rules (BR) to establishing a business and technical environment for managing BRs supported across the organization.
The UDFA process aims to decompose business logic into atomic-level BR’s. Using the Component-based Software Engineering model, the BRs are implemented as services, a concept that is very similar to data services provided by RDBMS. The Component-based Software Engineering model also provides for sub-processes, components that can invoke other components; providing a higher level of software re-use. Finally, the BR components are implemented by encapsulating them as user-defined functions (UDF), a relational-object extension of SQL. This creates a central repository of executables that can be invoked through SQL by any application running on any platform.
UDFA has great appeal to organizations seeking to managing automated BRs for the following reasons:
- No new hardware or software is required.
- Today’s application developer’s current skill set is sufficient to do the job.
- UDFA allows BRs to be implemented incrementally into legacy systems.
- The business people acquire management control and responsibility of their BR’s.
- The re-use of BR software provides faster, less expensive application development, testing and maintenance.
The road to success at NYSLRS began by presenting the UDFA concept and potential benefits to management. In response to management’s request for a proof-of-concept, the IT staff selected the Pension Payroll Calculator (PCC). The PPC computes zero to gross to taxable gross to net pension amount for an individual and is used primarily in producing the monthly benefit payments for 350,000 pensioners. The following statistics reflect the successful conversion of the legacy PPC to a BRM solution.
- The original block of embedded logic contained 1200 lines of source code.
- 17 BRs were harvested from this code.
- The 1200 lines of original source code were then replaced in the application with approximately 80 lines of code that fire individual rules and retain the results. This represented a 90% reduction in application code.
- 110+ lines of code were required to implement all 17 rules as separate UDFs.
- A sub-process was written that returns the next net monthly benefit for a pensioner’s id. This UDF demonstrated the powerful ad hoc query capabilities realized when BRs are part of the SQL engine.
- The sub-process was used to verify the BRM implementation of the PPC by running an ad hoc query after the monthly pension payroll was produced. The results returned by the query were identical to the net pension amounts shown in the payroll.
Next, the Benefit Projection Model Project was selected as a pilot project. This project consisted of 3 phases with the application targeted for 3 platforms; IBM mainframe. Lotus Notes Server and Java server. An additional requirement was that all computations for all three implementations take place on a single platform to ensure consistent results. The project kicked off with design meetings between IT staff and business people. During these meetings the necessary atomic-level BR’s and sub-processes were defined.
A key factor in end-user acceptance was Test-o-matic an inactive testing platform that was developed on a Lotus Notes server. The UDFs were tested by business people using Test-o-matic and building business confidence in the rule execution. This ability to test each UDF independently proved to the business people that each BR has a single physical implementation capable of being used by multiple applications. The ability to test easily and completely the business logic outside of the application was regarded by the business people as a great bonus.
It wasn’t long before UDFA was declared a complete success; a success that exceeded expectations. Changes were implemented in the day-to-day workflow to accommodate the full BRM process, from business inception to testing to production. Test-o-matic was expanded to serve as the source rule repository containing the documentation of each source BR, providing technical documentation of the corresponding UDF, automating the requests for registering a UDF in the DB, promoting a UDF to production, and requesting security be granted for the UDF.
The vision of turning a legacy system into a modern, agile system where the change process is accelerated by BRM is often tainted by the perceived need to acquire new hardware, software and/or staff skills, followed by the redesign and implementation of the system in a new operating environment. UDFA allows an organization to realize a true BRM vision by incrementally implementing BR’s into current systems. This delivers a pay as you go plan where the cost is minimal and the benefits often exceed expectations. If this approach seems too simple or too good to be true, keep in mind that IBM has endorsed this approach as a best practice for implementing BR’s into IT systems.
The most important conclusion from this experience is that BRM (and KPI RMM™ Level 2) can be achieved without significant financial investment and can even eliminate the need to build a business case for going forward. Putting together a business case for moving to BRM can be a significant undertaking; justifying culture change, new hardware ( BRE), and software platform. This article proves that UDFA can be used early as a low cost/low risk backdoor way to introduce BRM and with the right planning can be the beginning of a more strategic BRM vision. At NYSLRS we are looking to replace our 20 year old legacy system. The use of UDFA has eliminated the need for a BRM business case because BRM has simply become a serious requirement for the new system. In the mean time, we now know that we need to harvest rules in preparation for the replacement system and we know how to do that, but we can also implement the new rules into the existing system in the interim.
