David McGoveran

Work on applications of mathematical logic has pervaded Mr. McGoveran's career (1971–present). He has done original research and published on the structure of paradoxes,[7] applications of quantum logic to schizophrenia,[8] linguistic logic[9] and computational semantics (under James D. McCawley), fuzzy logic,[10][11][12] and applications of logic, including multi-valued logics, to databases.[13][14][15][16][17][18]

Beginning in 1981,[19][20] Mr. McGoveran began consulting on the design of transaction processing systems, including distributed transactions. Investigations into the complexity and cost of distributed transactions, as well as the difficulty of maintaining transactional consistency in online applications led to research into alternatives to the traditional transaction models that used pessimistic concurrency control and enforced ACID properties.[21] McGoveran defined physical transactions as the unit of recovery, logical transactions as the unit of consistency, and business transactions as the unit of audit[22] The resulting adaptive transaction model introduces a transaction intrinsic definition of consistency, deferring the decision to combine the results of two or more transactions. His work on transaction management resulted in the award of US Patent No 7,103,597.[23]

McGoveran's research on E.F. Codd's relational model has focused on the issues of data modeling (database design), missing information, and view updating. The last two are considered by some database researchers to be the most difficult and controversial problems in relational database research.[24]

Having worked on the design and development of several early large scale, distributed, commercial relational database applications,[25][26] McGoveran sought to improve upon the science of database design. This work lead to the development of

1. new analyses of and solutions to the problem of "missing information" and avoiding the use of nulls and therefore many-valued logic
2. the specification and uses of relation predicates (relation or set membership functions) as an application of Leibniz' Law[27][28][29]
3. a new design principle (with C. J. Date) now known as the Principle of Orthogonal Design (POOD)

His work on logic applied to relational databases and on design without nulls (1993) has been republished several times.[30][31]

McGoveran tackled the problem of view updating with Christopher J. Date starting in 1993 after having developed methods for reversible schema migration for clients on Wall Street.[32] His solution, based on relation predicates, formed the basis for the algorithms found in The Third Manifesto (Christopher J. Date, Hugh Darwen) for updating virtual relations (e.g., views). Date has credited McGoveran with originally suggesting the basic idea for the view updating approach,[33] and which Hugh Darwen[34] says represented a major shift in thinking on the issue. This work has resulted in two patents (U.S. Patent 7,620,664 and U.S. Patent 7,263,512).

Some of McGoveran's work on databases is discussed at Fabian Pascal's Database Debunkings web site.[35]

After consulting on numerous data integration and enterprise application integration projects, and related middleware products, McGoveran recognized that process aspects of integration were largely overlooked.[36][37] Most business process technology focused on analyzing and documenting existing business processes, then manually "reengineering" the processes to eliminate waste, remove bottlenecks, and improve cycle times. These efforts were largely disjoint from process automation systems and distributed control systems (which focused on highly repetitive, often continuous processes), and workflow technologies (which focused on highly repetitive sequential processes like document processing).

McGoveran postulated[38] an analogy between data management and process management.[39] Just as the relational data model proposed separating the logical model of the data from the physical storage model, it seemed that a logical process model (i.e., the business process model) should be separated from its physical implementation (e.g., as messaging, remote invocation, services, etc.). As with the relational model, this would permit business process design via models that were logically separated from specifics of process implementation, process scheduling, and process optimization. By introducing process measurement and analytics into the proposed process management system, closed loop process control became theoretically possible. The result was a set of requirements and a canonical architecture for the then largely unknown business process management system (BPMS).

The first commercial package compliant with this BPMS architecture ChangEngine - was then built and introduced by Hewlett-Packard in 1997-98 under McGoveran's direction.[40] Subsequently, McGoveran introduced these concepts at DCI's EAI conference in 1999,[41] through work as Sr. Technical Editor of the eAI Journal (Thomas Communications)[42] and worked with companies like IBM, Vitria, Candle, Fuego, Savvion, and numerous others to help shape the market and the BPM category.[43] Many workflow and business process reengineering (BPR) companies joined in the effort, transforming themselves into BPM companies during the period 1999-2010.

• McGoveran, D., Date, C. J. (1992). A Guide to SYBASE and SQL Server. Reading, MA: Addison-Wesley. ISBN 0-201-55710-X.
• Date, C.J., Darwen, H., McGoveran, D. (1998). Relational Database Writings, 1994-1997. Reading, MA: Addison-Wesley. ISBN 0-201-39814-1.

• McGoveran, D., (1991). The Evaluation of Optimizers. Encyclopedia Computer Science and Technology: Volume 26, Supplement 11. New York, NY:Marcel Dekker.
• McGoveran, D., (1993). The Evaluation of Optimizers. Encyclopedia of Microcomputers: Volume 13. New York, NY: CRC Press. ISBN 978-0824727116 & ISBN 0824727118.