The traditional wisdom for designing database schemas is to use a design tool (typically based on a UML or E-R model) to construct an initial data model for one's data. When one is satisfied with the result, the tool will automatically construct a collection of 3rd normal form relations for the model. Then applications are coded against this relational schema. When business circumstances change (as they do frequently) one should run the tool again to produce a new data model and a new resulting collection of tables. The new schema is populated from the old schema, and the applications are altered to work on the new schema, using relational views whenever possible to ease the migration. In this way, the database remains in 3rd normal form, which represents a “good” schema, as defined by DBMS researchers. “In the wild”, schemas often change once a quarter or more often, and the traditional wisdom is to repeat the above exercise for each alteration. In this paper we report that the traditional wisdom appears to be rarely-to-never followed for large, multi-department applications. Instead DBAs appear to attempt to minimize application maintenance (and hence schema changes) instead of maximizing schema quality. This leads to schemas which quickly diverge from E-R or UML models and actual database semantics tend to drift farther and farther from 3rd normal form. We term this divergence of reality from 3rd normal form principles database decay. Obviously, this is a very undesirable state of affairs, and should be avoided if possible. The paper continues with tactics to slow down database decay. We argue that the traditional development methodology, that of coding applications in ODBC or JDBC, is at least partly to blame for decay. Hence, we propose an alternate methodology that should be more resilient to decay.

There is considerable interest in moving DBMS applications from inside enterprise data centers to the cloud, both to reduce cost and to increase flexibility and elasticity. Some of these applications are "green field" projects (i.e., new applications), others are existing legacy systems that must be migrated to the cloud. In another dimension, some are decision support applications while others are update-oriented. In this paper, we discuss the technical and political challenges that these various enterprise applications face when considering cloud deployment. In addition, a requirement for quality-of-service (QoS) guarantees will generate additional disruptive issues. In some circumstances, achieving good DBMS performance on current cloud architectures and future hardware technologies will be non-trivial. In summary, there is a difficult road ahead for enterprise database applications.

Our world is increasingly data-driven. The growth and value of data continue to exceed all predictions. Potential for business opportunity, economic growth, scientific discovery, and even national security lie hidden in the vast and growing data collections distributed across our Digital Universe. Harvesting the value of data requires finding, integrating, and analyzing data distributed across our Digital Universe. Failing to integrate data could be as mild as losing business through a lack of an integrated view of relevant business opportunities and threats or as serious as not anticipating the most serious terrorist attack on America through an inability to integrate relevant information distributed across 17 US intelligence agencies and 28 intelligence databases. Due to data being distributed across both entities and data repositories, data integration has become perhaps the most important data operation but certainly the most costly accounting for up to 40% of data processing budgets. Data integration requirements are growing dramatically due to astounding growth in semantics (i.e., increasingly sophisticated application and data modeling requirements) and scale (i.e., web-scale data and transaction types, volumes, and distribution). Since the mid-1980's the Relational Data Model has been the bedrock of business data processing providing a simple, elegant, and powerful basis for storing, finding, and integrating data based on simplifying assumptions. First, business data is naturally tabular; second, tabular business data follows relational principles; and third, assuming that there is a single source of truth, data views can be integrated into a single view under a global schema and that views of the same information are mutually consistent. To the extent that application data and operations fit naturally and meaningfully into the relational model and follow the three assumptions, Relational Database Management Systems (RDBMSs) provide the most efficient and cost effective data integration solution on the planet and will continue to do so through hardware and engineering advances. As semantic and scale requirements of business data grew in the 1990's relational integration had to be extended ideally internal to the relational database engine, otherwise externally in data integration tools. This led to a plethora of data integration tools and supporting infrastructures with sales exceeding $3.3 billion in 2009 with an expected annual grow rate of 8%. These integration solutions are typically less efficient than relational integration and considerably more costly to acquire and more complex to deploy, yet respond to genuine, if esoteric, data integration requirements. While relational integration in RDBMSs augmented by integration tools meet the requirements of many large-scale applications, they are increasingly less applicable to a growing class of very-large-scale Information Ecosystems that can involve thousands of information systems and databases. Most Fortune 500 companies have ten or more major organizations Corporate Management, Sales, Marketing, Engineering, Product Development, etc. each requiring it's own view of the enterprise from it's unique perspective. Such views are often derived (i.e., integrated) from 5,000-10,000 databases. The Information Ecosystem is the information systems, databases, workflows, people, and infrastructure required to build, maintain, and dynamically update (i.e., integrate) the myriad organizational views. While the relational model and some combination of the three simplifying assumptions may apply to each component information system or database, the Information Ecosystem does not. While an individual information system may be designed to simplify the semantics of the real world application that it represents, the complexity of the real world or of the interactions between 1,000's of information systems cannot be as readily simplified. That is, relational integration and integration tools may apply to an individual or to a small number of information systems, but they are less applicable to large numbers of information systems such as occur in Information Ecosystems. Relational technology is one of the most successful computing technologies in history based in part on the elegance, simplicity, and power of the Relational Data Model. It established a prototype for the next thirty years of database developments. Until the .comand Web-fuelled expansion of our Digital Universe, the majority of data intensive applications were built on the bedrock of relational technology. Applications that did not fit or could not be forced into relational databases were built as extensions to relational technology or as entirely separate solutions. Both methods were not successful such as the rise and fall of objectoriented databases. It now appears that the Relational Data Universe is less than 15% of the Digital Universe. Yet harvesting the value of data in our Digital Universe still requires finding, integrating, and analyzing data distributed across the Digital Universe. If relational integration applies to less than 15%, what data integration solutions are required for the other 85%? The rapid expansion of our Digital Universe both by amazing applications spawned by the Web and by phenomenal advances in hardware speeds and costs led to a continuous stream of semantic and scale requirements that far outpaced data integration solutions and has spawned or been spawned by new technologies such as Web search, Collaborative Filtering, MapReduce, the Semantic Web, Ontologies, Cloud computing, and a dizzying array of others. This talk surveys the evolution of data integration requirements and solutions from the Relational Data Universe to the Digital Universe. It examines data integration solutions in the Relational Data Universe starting with the relational data model and the three simplifying assumptions. It considers the major advances in data integration solutions in terms of semantics; database engine capabilities including data format, data architecture, and data governance; and the integration methods of integration tools. We consider a very-large-scale Information Ecosystem to illustrate limits of data integration solutions in the Relational Data Universe and conclude with review of directions towards data integration beyond the Relational Data Universe including new data models, a profound new direction from the Database Community; Information Ecosystems; Semantic Technologies emerging from the Semantic Web community; and Cloud computing where data integration may reside as Integration-as-a-Service or Informationas-a-Service (IaaS either way!).

Summary form only given. This presentation is an exploration of the next generation computing based on the confluence of radical and coincidental changes in economics, business, and technology. Whereas technology is a key enabler of change, it is and is the servant, not the master. Without a depth of understanding of this enabling role and the content in which technology serves, technology can be misguided and its developers can lose perspective. This presentation outlines a proposal made to the US President’s Office of Science and Technology for technology research for the next decade, which calls for new computational models, operating systems, data models, and infrastructure, amongst others, to support the next generation of computing, collaborative computing. This provides an opportunity to reconsider the role of data in computing. This is only one view.

Future information systems will involve large numbers of heterogeneous, intelligent agents distributed over large computer/communication networks. Agents may be humans, humans interacting with computers, humans working with computer support, and computer systems performing tasks without human intervention. We call such systems Intelligent and Cooperative Information Systems (ICISs). Although we can imagine extensions of capabilities of current ISs and of individual contributing core technologies, such as databases, artificial intelligence, operating systems, and programming languages, we cannot imagine the capabilities of ICISs which we believe will be based on extensions of these and other technologies. Neither do we know exactly what technologies and capabilities will be required, what challenges will arise, nor how the technologies might be integrated or work together to address the challenges. In this paper, we provide initial definitions for key concepts and terms in this new area, identify potential core contributing technologies, illustrate the ICIS concept with example systems, and pose basic research questions. We also describe the results of discussions on these topics that took place at the Second International Workshop on Intelligent and Cooperative Information Systems held in Como, Italy, October 1991. The workshop focused on core technologies for ICISs. The workshop and the results reflect the multi-disciplinary nature of this omerging area.

Future information processing environments will consist of a vast network of heterogeneous, autonomous, and distributed computing resources, including computers (from mainframe to personal), information-intensive applications, and data (files and databases). A key challenge in this environment is providing capabilities for combining this varied collection of resources into an integrated distributed system, allowing resources to be flexibly combined, and their activities coordinated, to address challenging new information processing requirements. In this paper, we describe the concept of distributed object management, and identify its role in the development of these open, interoperable systems. We identify the key aspects of system architectures supporting distributed object management, and describe specific elements of a distributed object management system being developed at GTE Laboratories.