Incremental Interoperability and an Industrial Revolution: Why Data Interoperability Fails, and How We Can Do Better

Arnon Rosenthal (with Frank Manola, Scott Renner)
The MITRE Corporation

arnie@mitre.org

Abstract

Many initiatives, governmental and commercial, have pursued the grand vision of "transparent access" - making all data available to all consumers (users and applications), in a way the consumer can interpret, anywhere and at any time. Among large-scale enterprises, success stories in achieving such visions seem rare or nonexistent. Instead of leading one more charge "over the top", we suggest that a better guiding metaphor would be "continual evolution of a partially satisfactory system". This resembles what we have today, and what we will have in 2010. We first examine fallacies that need to be corrected if there is to be a hope of success. We then outline an "industrial revolution" that partitions work into simpler, one-skill pieces. Many pieces will be automatable; others could be done by software-naïve end users; all produce reusable knowledge. We identify first steps that may offer some immediate benefits, give crucial stakeholders (metadata providers) incentives to participate, and yet fit a long term vision of greatly increased access. We also describe open research problems in data administration and query brokering.

Key Words - data integration, metadata, enterprise systems, interoperability, architectures

Biography

Dr. Arnon Rosenthal works in many areas of data management research, and on connections to object management. A Principal Scientist at The MITRE Corporation, he is involved with advanced data management projects throughout the company. His current research seeks to simplify administration of data shared among many user communities, including data warehouses, federated databases, and business object suites. One line of research examines security for such environments; another addresses frameworks for general metadata administration there. Previous work (over the years) has included "annotations", interoperability, distributed object management, migration of legacy systems, and query processing. Previous employers include Computer Corporation of America, ETH Zurich (visitor), Sperry Research, and U. Michigan. He holds a PhD from U. California (Berkeley) .

For mere details, see: http://www.mitre.org/resources/centers/it/staffpages/arnie/

PS: MITRE is hiring. Email me your resume.