|
Management of enterprise-wide databases and access to managed
data is a growing challenge for data processing customers.
Databases that have grown over time within various parts of an
organization may not share a single view of the data, may have
different tools, and may lack consistency in format and meaning.
In addition, these large databases may lack a record of changes,
may have conflicting rules for data sharing and general data
administration, and may present a complex web of management
issues such as timeliness, support, integrity, and maintenance.
This issue presents a current view of the technologies and
environments for management of large enterprise databases. Seven
papers show current experiences and solutions, extensions to the
relational model and systems, support for object management
systems, distributed databases, and parallel processing for
database systems. The experience, knowledge, and technology
described is, in turn, the focus of the ongoing development of
IBM's Information Warehouse* framework. This framework supports
relational and object approaches to data and encompasses an
architecture and conforming implementations. We are indebted to
N. E. Owens of the IBM Santa Teresa Laboratory, IBM Software
Solutions Division, in San Jose, California, for her coordination
and management of the preparation of these papers.
Some enterprises are addressing the current complexities of
enterprise-wide databases and data by applying innovative
approaches to their existing systems and technology. In the
first of two papers addressing current solutions,
Bustamente and
Sorenson describe the management of operational data gathered
from order processing and customer mailing at Lands' End, Inc.
The resulting corporate database is accessed through a decision
support system built using DATABASE 2* (DB2*), in a way that can
be considered a precursor to and subset of the Information
Warehouse framework. The paper discusses corporate requirements,
database design process and modeling, tool selection, and
insights gained during implementation. (Reprint Order No.
G321-5540)
Schlatter et al.
offer a second paper on current experiences
that also treats technical advances intended to pave the way
for enterprise-wide database solutions. Their Business Object
Management System (BOMS), which is operational today, is intended
to support billions of objects of all kinds at one time. It is
built on DB2 and Customer Information Control System/Enterprise
Systems Architecture (CICS/ESA*), extending the relational
database model in ways that avoid the problem of predicting
future uses of data and relationships among data. It also
provides transparent access to data, portending a future of
cooperative database management in which data navigation is no
longer an end-user concern. (Reprint Order No. G321-5541)
Turning our attention to technology for the Information Warehouse
framework, both a bridge for existing database management systems
and support for newer approaches are needed, so that the current
and foreseen challenges for databases can be met. The first
paper of two on technology presents extensions to relational
database management systems (RDBMSs) and models, supporting new
capabilities such as extensible types and functions, a global
semantic rules subsystem, and performance enhancements. These
extensions, as described by Cheng et al.,
are currently
implemented and tested as prototypes for the Structured Query
Language (SQL). (Reprint Order No. G321-5542)
In the second paper of this set, Alfred
provides a view of the
technology for ObjectStore**, an object database management
system (ODBMS) that is now part of IBM's object database
solution and the database component of IBM's FlowMark* work flow
management system. He discusses two paradigm shifts and the
leverage they provide: the object-oriented development model and
the direct-reference storage model. These models underlie some
current database systems, including ObjectStore, and are part of
the adoption of increasingly sophisticated object-based software
development techniques. (Reprint Order No. G321-5543)
A further set of two papers centers on the design and development
of distributed database applications within the context of the
Information Warehouse framework. The first, by
Singleton and
Schwartz, explores direct data access in a heterogeneous systems
environment, with the goal of supporting efficient database and
systems administration for a large enterprise. Specific software
development techniques are presented, along with technical
trade-offs that need to be considered in their use. (Reprint
Order No. G321-5544)
Leymann and Altenhuber
provide a second paper on distributed
databases that focuses on the modeling, management, storage,
and execution of enterprise-wide business process descriptions,
and their importance as information assets. The meta-model
supporting this work utilizes weighted, colored, directed graphs
of activities, executed by graph navigation. This meta-model is
implemented in FlowMark and used for the work flow management
component of business processes. Comparisons are made with other
methods, such as Petri nets and process models. (Reprint Order
No. G321-5545)
The pressure for efficiency and rapid responses to queries in
large database systems has made it beneficial to consider
parallelism for query processing. With a proposal for using
parallelism comes the need to consider secondary characteristics,
such as load balancing, pipelining, database design, and
parallelization of the query itself. The authors,
Mohan et al.,
show how these questions were resolved and implemented in the
chosen relational database management system: DB2 Version 3.
(Reprint Order No. G321-5546)
The next issue of the Journal
will be a special issue on projects
sponsored by IBM Canada's Centre for Advanced Studies.
Gene F. Hoffnagle
Editor
(C) Copyright 1994 by International Business Machines Corporation.
*Trademark or registered trademark of International Business Machines Corporation.
**Trademark or registered trademark of Object Design, Inc.
|
HTML