Recession, Technology,
William H. Hinkle, Founder & Manager, N_gine LLC
Copyright N_gine LLC 2008
Introduction
One certain means of reducing the
frequency, severity, and duration of recession is to prove continuous superior value
propositions of products and services against the competition. This means reinventing the content and
reducing the cost of products and services across those industries addressing
basic human needs. Prime vertical market
examples include Automotive, Banking, Consumer Products, Energy, Financial
Services, Food, Government, Health Care, Insurance,
New Technologies
Two new technologies having profound, pervasive, and permanent effects on both content and costs of financial transaction processing are (1) parallel processing and (2) 10 Gigabit networking. Parallel processing is the ability to utilize two or more processors to solve a common problem. Parallel processing has existed for several decades in scientific applications that are computationally intensive. However, parallel processing has yet to emerge for business applications that are storage intensive. 10 Gigabit networking is simply a ten-fold increase in present networking speeds, which suggests more access to essential financial information in real-time. The latter is merely a matter of installation whereas the former requires total rethinking of software architecture for business applications.
Parameterization
The subject of parameterization deserves mention before describing parallel processing. The reason being: Parameterization (or user-definability) enables users to define their own respective financial transaction processing environments via data rather than code. Such definitions include financial reports, financial transactions, and financial processes. Thus, nuances of the same application and integration of disparate applications may be combined within the same software system at the same time. Proper parameterization means substantially shrinking the required code, which in turn shrinks the number of programmers that in turn shrinks the costs of evolving software development and maintenance. Parameterization also extends solution life cycles indefinitely and enables parallel processing.
Processing Modes
Presently, virtually all business software processes financial transactions serially, one transaction after another. If one financial transaction updates more than one file, then all files are updated one after another in serial fashion. In contrast, parallel processing means that all file updates occur simultaneously. Therefore, each file update must be separate and independent of all other file updates. Otherwise, when one file update depends upon another, processing reverts back to serial processing in some degree. Parallel processing shortens response times, an important benefit for real-time financial decision-makers such as securities trading. However, there is one enormous problem.
Technical Problem
Parallel processing requires decomposition (or division) of total job functionality into a set of small, separate, and independent sub-processes. Decomposition must occur at the lowest possible atomic level so that as many processors as possible may be assigned to reducing response times. Likewise, all resulting sub-processes must also be of the same processing duration to ensure proper “load balancing” across all processors. Otherwise, the sub-process with the longest duration will determine the parallel processing benefit. Needless to say, decomposition into smaller sub-processes is highly complex for even those most familiar with the jobs to be processed. Thus, the growth rate of parallel processing will be constrained by the number of persons with application knowledge and decomposition skills rather than technical knowledge and programming skills.
Innovation Sponsors
Intel and Microsoft legitimized
the parallel processing problem in Q1-2008 by granting $20 million ($10 million
each) to the
Parallel Processing
True parallel processing means that an unbounded number of processors are possible in financial transaction processing, where each processor supports at least one file that may be used separately or jointly with other files in creating any number of financial decision-making reports. Some simple financial transactions may require five simultaneous processors whereas some complex financial transactions may require thirteen processors. Some financial transactions could even update fifty or more financial reports.
Specific Breakthrough
The point is: Parallel processing supports an infinite number of real-time financial reports simultaneously that are concurrent across an entire enterprise. This is the content mentioned at the outset of this article – all relevant financial reports are immediately updated and accessible after each financial transaction is entered. Simultaneous content means concurrency of all financial reporting for better decision-making, where most all batch processing is shifted from overnight to real-time. Hence, totally new software architecture for core back-office financial transaction processing is required.
Additional Resolutions
Unfortunately, new 21st century financial transaction processing software must also remedy 20th century legacy software problems of bloated code, memory thrashing, disk contention, and downstream systems. The problems with bloated code are: millions of lines of code exist, documentation is typically poor, patches are expensive, and personnel are in short supply. The problem with memory thrashing is that 20th century software is characterized by few large programs that exceed the size of available memory and various code segments have to be constantly reloaded into memory to process each transaction. Such reloading is called “memory thrashing” (or “machine overhead”), which is probably today’s biggest waste of time and electricity. Disk contention results from many users requesting the same data stored on disk, where some users have to wait for others to complete their jobs before being serviced. Many downstream systems result from incoming data having to be reread several times to complete several different jobs when one read should have been sufficient.
First Solution
N_gine LLC (pronounced “engine”) of Denver, Colorado has created, patented, and tested new real-time parameterized parallel processing of financial transactions that also minimizes if not eliminates the residual 20th-century problems. N_gine proved 28,000 financial transactions in less than five minutes and 280,000 financial transactions in less than fifty minutes on a quad-processor machine costing less than $10,000. Each financial transaction updates all relevant financial reports including Balance Sheet, Income Statement, Cash Flow, Performance Measurement, Capital Gains, Pending Income/Expenses, Pending Corporate Adjustments, General Ledger, and Transaction Journal. The financial transactions included Debits, Credits, Deposits, Withdraws, Contributions, Distributions, Buys, Sells, Exchange-In, Exchange-Out, etc. More financial reports, financial transactions, and financial processes may be added with relative ease. Such architecture may be in its infancy, but the risks of technical viability, operational scalability, and economic desirability have already been minimized.
New Tests
N_gine has advanced production to 44,000 financial transactions in less than five minutes by utilizing the fastest disk controller and disk drives (15K RPM) in the marketplace. N_gine is now planning tests on blade servers with fourteen blades, where each blade server may be perceived as a laptop with its own separate processor and dedicated disk drive. These machines are expected to introduce another 5 to 10-fold productivity increase. Subsequently, N_gine intends to test on mid-range and high performance mainframes where no productivity estimates have yet been made. Accordingly, N_gine seeks sponsors to participate in the testing and receive financial favor in licensing the results. Parameterization and parallel processing suggest entirely new financial transaction processing content and cost structures. Much of N_gine is patented. Further intellectual property efforts will continue.
Technical Surprise
The biggest surprise in N_gine development requires explanation. In over-simplified terms, a computing solution consists of four levels: hardware, operating system, relational database system, and application. The hardware consists of processor, memory and disk drives. The operating system is Microsoft Windows XP, UNIX, LINUX, etc. The relational database is Oracle or IBM’s DB2. N_gine is the application, where the user selects either the Oracle 10g relational database system version or the IBM’s DB2 relational database version. Both have been proved. The surprise is: The total executable code for either version (the operating system code plus the relational database code and the N_gine code) will fit into 2 gigabytes of memory (the amount of memory found in most high-end laptops). This means that “memory thrashing” or “machine overhead” is virtually eliminated and full computing power is applied to process each transaction. As an Independent Software Developer, N_gine LLC could not be more shocked and pleased. Small is good! Small saves money.
Marketing Implications
N_gine addresses the Financial Services Industry, Governmental Agencies, Charities, the Fortune 2000, and others processing large volumes of daily transactions. Much like the transition from analog to digital television, the transition from serial to parallel processing will be profound, pervasive, and permanent. N_gine is ideally suited to hosted applications attached to global networks where charges are made on a per-use basis (Software as a Service). Parallel processing and 10 Gigabit networks will serve to provide global information in larger quantities at lower costs. The new total productivity is unprecedented. Fast is good! Fast saves time.
Corporate Implications
Parameterization and parallel processing have the potential to alter business models, corporate strategies, organization charts, marketing plans, and operating procedures. Macroeconomics defines a production function as the summation of all costs in creating products and services and the support thereof. For Information Technology, these include equipment, personnel, space, electricity, administration, and top management. Parameterization and parallel processing affect each cost and all costs collectively – fixed costs and variable costs.
Productivity Goals
N_gine’s technical goals target 50-90% code reduction, 5 to 10-fold productivity increases, and 30% greater ROIs than present automated investment returns. Parameterization targets the first, where one software system is capable of processing two or more disparate financial applications. Parallel processing targets the second, where fourteen processors will be used in processing a single financial transaction on blade server machines. Simulation models are used to test the third. If all criteria are met, then pilot projects are begun where accuracy, performance, and capacity, are constantly monitored on a daily basis.
Management Role
Management’s role is to define future competitive advantage (content) and operating constraints (cost). Yet, most top management is not yet aware of the power of parameterization and parallel processing. Top management must accept the responsibility of sufficient education to compete. Innovation, education, and leadership are necessary to restore confidence in the economy.
Today’s Opportunities
Today’s opportunities are unbounded. Tomorrow’s leaders will be the first to change. Parameterization is not only our best hope but our only hope for increased content. Parallel processing is not only our best hope but our only hope for lower costs. Together, they represent the best hope for global financial transaction processing. N_gine offers conceptual presentations and live demos NOW! Bank Holding Companies are now receiving funds from the Troubled Asset Recovery Program (TARP). TARP will require clarity, transparency, auditability at low costs. Therefore, any funds earmarked for future investment in automation should focus on the value propositions of 21st century innovation. Innovation is good! Invest in innovation.
Acquisition Questions
Seven questions to ask when acquiring new financial transaction processing software are:
(1) Does the software meet you financial decision-making needs?
(2) Is the financial transaction processing environment totally parameterized?
(3) Does all financial transaction processing occur in true parallel mode?
(4) Are all changes to the financial transaction processing environment fully auditable?
(5) Is the total financial transaction processing environment portable across machines?
(6) What is the cost of software acquisition, operation, support, and evolution?
(7) Are savings in time and money readily identifiable and predictable?
Implementation Plan
N_gine’s implementation plan is as follows:
(1) Plan integration of disparate financial applications from the outset.
(2) Parameterize the total financial transaction processing environment in terms of financial reports that satisfy the financial decision-making needs, financial transactions that update the financial reports, and financial processes that comprise each different type of financial transaction.
(3) Develop a meaningful set of test data, with multiple datasets to test capacity.
(4) Define different simultaneous production lines if needed.
(5) Decompose every financial transaction to its lowest possible atomic level to ensure the fastest possible response times and load balancing across all processors – all decompositions containing one and only one disk operation..
(6) Capture every relevant transaction (or change) in every organization every day.
(7) Track every transaction (either originating or reversing) from point of inception (data entry) to permanent archive (data warehouse) without overwriting any original data.
(8) Allocate each business process to a separate and independent processor.
(9) Execute all processors simultaneously under full audit control – no partial transaction processing update being allowed.
(10) Process every transaction to its logical conclusion to minimize if not eliminate any downstream systems.
(11) Audit the entire financial transaction processing environment on a frequent basis to ensure system integrity.
(12) Repeat items 6 – 11 until incoming data stream is exhausted.
(13) Provide complete Backup and Recovery of all data for off-site processing.
(14) Provide Production, Backup, and Development systems for total operations.
(15) Secure operations with latest methods.
(16) Provide completeness, clarity, and transparency.
Auditing
Such general steps are detailed in the Federal Financial Institution Examination (FFIEC) Council manuals and used by the Office of the Comptroller of the Currency (OCC) and the Securities and Exchange Commission (SEC).
Future
N_gine LLC is an outsourced Research and Development resource for global companies seeking new profits in processing large volumes of daily financial transactions. N_gine targets parameterization, decomposition, parallel processing, and 10 gigabit networks. N_gine charges one-time license fees, on-going support fees, consulting fees, training fees, storage fees, and archiving fees. Accordingly, N_gine seeks five initial users at $2 million each to support a staff of 30 persons, equipment, etc, where the five initial participants would receive 20% of all future revenues until their original investment is recovered.
Contact
Mr. Bill Hinkle, Founder & Manager Telephone: (303) 995-7675
N_gine LLC Email: billhinkle@nginert.com