The Case for New Back-office Automated Systems for Financial Institutions
The case for replacing present back-office systems of financial institutions is really quite simple. Today’s serial processing systems cannot keep pace with rapid changes in consumer demand, regulatory compliance, global competition, and rapidly advancing technology. Increasing front-end access demands are causing larger overnight demands where transparency is required across a host of related activities. Operations continue to be fragmented. New regulations are widely anticipated. Automated systems must anticipate more changes, and simplicity is in demand.
Primary Driver
Yet, the real “bolt from the blue” is the paradigm shift in hardware from single core to multicore chips that dictates a subsequent paradigm shift in software from serial to parallel processing. In specific, the execution of serial programs on multicore chips is futile because the serial code contains no logic for utilizing the incremental chip(s). Many are now recognizing the problem.
Research Efforts
During Q2-2005, Intel, IBM, and
AMD announced multicore chips, where two or more processors reside on a single
chip creating parallel processing potential.
In Q1-2008, Intel and Microsoft granted $20 million ($10 million each)
to the
Immediate Need
What is immediately needed is a clear picture of what the future looks like and how to get there. New concepts, architecture, technology, and education are required if new results are to be attained. The purpose of this article is to describe two concepts proven to have the greatest impacts on future financial transaction processing profitability: Parameterization (the “what”) and parallel processing (the “how”).
Parameterization (User-Definability)
Parameterization is a fancy term for enabling users to define (or customize) their own respective financial transaction processing environments in terms of financial reports, financial transactions, and financial processes. No two nuances of the same application have to look or act alike. All definitions are loaded to the software as data rather than code. There is no reason to maintain two or more versions of the code. Thus, parameterization suggests utilizing one software system for processing two or more disparate financial applications and spreading the cost across the two or more applications. This transformation of logic from code to data is the most important contributor to three-digit increases in productivity and low costs.
Parameterization also reduces the size of the executable code to fit into memory, thereby minimizing if not eliminating memory thrashing. The result is faster responses, greater machine capacity, and lower electrical cost. Most importantly, parameterization of the actual financial transaction processing logic creates the links to parallel processing.
Parallel Processing (Simultaneous Execution)
Parallel processing means the simultaneous execution of two or more jobs. The concept of parallel processing is not new. Parallel processing has existed in computationally-intensive scientific and graphics applications for decades. However, the parallel processing of storage intensive business application remains elusive because of the serial disk updating and subsequent disk contention.
Processing the disk drives in parallel is the key to unprecedented productivity never possible with serial processing. All relevant financial reports are updated after each financial transaction is entered - creating complete, consistent, and real-time reporting. Each disk operation is defined as separate and independent of all other disk operations to achieve the maximum benefit of parallel processing. Otherwise, the dependency of one disk operation on another disk operation causes processing to fall back to serial mode.
Success Criteria
N_gine LLC is first to emerge with broad parallel processing capabilities based on proper financial transaction processing success criteria. Proper success criteria (1) captures every relevant change (or transaction) in every organization every day, (2) tracks every change (either originating or reversing) from point of entry (data entry) to permanent archive (data warehouse) without ever overwriting any original data, (3) processes every transaction to its logical conclusion so as to minimize downstream systems, and (4) performs backup/recovery for all data before ever considering parallel processing.
Financial Objectives
N_gine targets reducing the code by 50-90% to increase productivity by 5 to 10-fold, and generate 30% greater ROIs than present automation alternatives. In short, N_gine has miniaturized the code while increasing content, speed, flexibility, reliability, and profitability.
Basic Procedures
The four basic procedures for realizing real-time parameterized parallel processing of financial transaction processing are:
(1) Parameterize the total financial transaction processing environment including relevant financial reports satisfying the financial decision-making need, financial transactions that updating the financial reports, and financial processes that comprising each type of financial transaction.
(2) Decompose the each type of incoming financial transaction into a set of small, separate, independent financial processes that may be spread across any number of separate and independent processors. All financial processes should be decomposed to their lowest possible atomic level to employ the greatest number of processors and achieve the fastest possible response times. And, all financial processes must be of the same processing duration to achieve proper load balancing across all processors. Otherwise, the financial process with the longest processing duration will determine the productivity of parallel processing. Then, assign the appropriate input data into the appropriate processes.
(3) Allocate all separate financial processes across the separate processors of same processing speed, each financial process being immediately executed.
(4) Execute all financial processes in overlapping fashion under full audit control.
While each of these topics could be subject of a doctoral dissertation, each may be explained in simple terms. N_gine is a “user-definable high-volume financial transaction processor” capable of satisfying many disparate financial transaction processing needs. User-definability (or parameterization) shrinks the actual code required by 50-90%. What code remains minimizes memory thrashing and processes disk operations in parallel to achieve the fastest possible responses. Legacy serial processing cannot and will never match the benefits. The new software architecture is independent of most all financial applications, relational database, operating system, and vendor hardware.
Technical Focus
Prospective users of parameterized parallel processing software will note less emphasis on processor speeds, the number of processors per chip, and memory speeds and more emphasis on system configurations, fast disk drives, network speeds, and security. Users will also note that N_gine unique system configuration is portable across quad-processors, blade servers, mid-range machines, and mainframes. This means that N_gine is less expensive to “build out” or “build up” as loads grow and the economics of computing change.
Financial Focus
N_gine’s constant goal is to lower fixed costs and variable costs while producing greater profit margins and returns on investment. Migrations are best achieved by starting small and then growing to the desired machine in triplicate – production, backup, and development. Fortunately, small machines in triplicate require smaller budgets than mainframes. N_gine’s plan is to enable users to port operating system, database, application, and data easily from any one machine to any other machine.
Best Scenario
Today’s methods are not viable solutions for tomorrow. Tomorrow’s solutions are not high speed yesterdays. N_gine’s patented technology and parallel processing code represent a solid foundation upon which to build enduring solutions. Intel, Microsoft, UC-Berkeley, UI-Urbana/Champaign, and Stanford will unquestionably produce more efficient parallel processing platforms. N_gine is first to offer a first practical application to sit atop those platforms. Much collaboration is to be had in reaping the full parallel processing potential for the public good.
Worst Scenario
The worst possible scenario for global competitors is to be locked into long-term contracts employing over-bought machines running non-integratable bloated serial software that wastes untold electricity, incurs high maintenance costs, and risks covering monthly fixed costs during recessions. New low-cost 21st century hardware requires new low-cost 21st century software for unprecedented competitive advantage, stable profit margins, and greater returns on investment.
Today’s Profit Opportunity
Today’s profit opportunity is to invest in multicore computing to create multi-application solutions, where parameterization and parallel processing pave the way to new profitability. How profitable? Forbes Magazine reports Nasdaq OXM saving $100 million by replacing mainframes with commoditized processors. N_gine suggests that parameterization (smaller code) and parallel processing (overlapping disk operations) would swell profits several times more. Parameterization is not only our best hope but our only hope for competitive advantage. Parallel processing is not our best hope but our only hope for operating efficiency. Combining both is not only our best hope but our only hope for sustainable attractive returns on investment. Any other financial transaction processing system lacking either or both is destined for complete redesign from the ground up. Global competitors are watching. Case closed. Look to N_gine.
Contact: Bill Hinkle, Founder & Manager, N_gine LLC, 3524 S. Hillcrest Drive, Denver, CO 80237 Tel: (303) 995-7675; billhinkle@nginert.com