VLDB '95 Tutorials
Tutorial A
MO 11 09, 14:00 - 17:30
A Survey of Parallel
Database Techniques and Systems
Jim Gray (UC Berkeley)
Abstract
The tutorial first introduces the
common techniques used by parallel database systems to partition data and
execution. Parallel variants of the relational operators are described.
Optimization techniques are then briefly outlined. Attention then turns to
commercial systems available today. The tutorial highlights the strengths of
each of these systems: Teradata, Tandem, Oracle, Informix, Sybase, and DB2/2.
About the Instructor
Jim Gray is a specialist in
database, transaction processing, and dependable computer systems. He has
written many articles and built several systems that embody his ideas on these
topics. He is currently a McKay Fellow at UC Berkeley. He has worked at Digital
on Rdb and Sorting, worked at Tandem on NonStop SQL, and at IBM on System R,
SQL/DS, DB2, and IMS-Fast Path. He is co-author of Transaction Processing
Concepts and Techniques, and Editor of the Handbook for Database and Transaction
Processing Systems Performance. He is a Fellow of the ACM and is active in the
National Research Council.
Tutorial B
MO 11 09, 14:00 - 17:30
Spatial Database
Systems
Ralf Hartmut Gueting (University of Hagen)
Abstract
The purpose of a spatial database system is the management of data about
some space in two or more dimensions. Space of interest may be the physical
world around us as in geography, urban planning, meteorology, or astronomy.
Other spaces may, for example, represent parts of the human body, a VLSI design,
or a 3D molecular structure. A spatial database keeps information about objects
in such spaces with a clear identity, position, and extent, in contrast to image
databases where the focus is on the manipulation of raster images of such
spaces.
Spatial data management imposes a number of requirements on database
systems. Concepts are needed to describe the shape and position of objects in
space as well as relationships between them. Data modeling must also provide
concepts for the representation of spatially related collections of objects,
such as partitions (of a region), or networks. To access efficiently objects in
some particular region of the space, specialized spatial access methods are
needed. The user interface must support graphical representation as well as
overlay of different query results. Since extensions at all levels of the system
architecture are needed, extensible database systems are the prime architectural
framework for the implementation of spatial database systems.
The tutorial
aims at giving a coherent picture of the main research results obtained so far
in the areas of modeling, querying, data structures and algorithms for system
implementation, and system architecture.
About the
Instructor
Ralf Hartmut Gueting is currently a full professor in
computer science at the University of Hagen, Germany. He received his Diplom and
Dr.rer.nat. degrees from the University of Dortmund. After a one-year stay at
the IBM Almaden Research Center in 1985, extensible and spatial database systems
became his major research interests. His group has built a prototype of an
extensible spatial DBMS, the Gral-System. He is the author of a (German) text
book on data structures and algorithms and has published about 35 articles in
computational geometry and database systems. He recently was a guest editor of a
special issue of the VLDB Journal on Spatial Database Systems.
Tutorial C
TU 12 09, 14:00 - 17:30
The Evolution of User
Interface Tools for Database Applications
Ravi Krishnamurthy
(Hewlett-Packard Labs)
and
Moshe Zloof (Hewlett-Packard
Labs)
Abstract
Traditionally, the user interface tools for
database applications have lagged the technological advances in GUI revolution.
Recent entrants such as PowerBuilder, SQL Windows, NextStep, etc., have made
major strides in closing this gap. In order to extrapolate the trends, we
introspect the past and forecast the future.
We will first give a historical
account of the evolution of database UI tools since the early seventies in the
context of many facets of database applications (e.g., querying, forms &
reports, 4GL capabilities, and end user graphical tools). These tools were aimed
at improving the productivity of application developers, the primary clientele
of SQL.
The advent of PC database management system saw the metamorphism of
this clientele to less sophisticated users. These non-programmers (i.e., power
users) who wish to build their own applications required different kinds of
tools and the PC database industry lead this thrust.
We address the design
factors in both these types of tools and contrast them. In doing this, we will
use examples from some of the current products (e.g., Microsoft's Access,
Gupta's SQL Widows, Powersoft's PowerBuilder, etc.) to highlight their power,
functionality, levels of abstraction and ease of use.
Lastly, we extrapolate
on how this UI technology will be evolving in the future.
About the
Instructors
Ravi Krishnamurthy is a senior researcher at H.P. Labs,
currently heading the Picture Programming project that is developing the
IC-BY-EXAMPLE language. He received his doctorate degree at the Univ. of Texas
at Austin and then joined Moshe Zloof at IBM T.J. Watson Res. Center to work on
the Office-By-Example project in 1981. In 1985, he joined MCC to work on Budda
and LDL, where he subsequently assumed the chief architect position of the joint
project. After leaving MCC, he was a co-founder of a start-up company, Logica
Information Machines, in which he is still involved as its Chairman of the Board
of Directors.
Moshe Zloof, principal architect of H.P. Labs, is considered a
poineer researcher in the area of data base languages and user interfaces. Back
in the 1970's, in a complete departure from the traditional approach and while
working at IBM, Moshe created QUERY-BY-EXAMPLE (QBE), the first visual
programming language which not only set the stage for considerable research
agenda, but also has been incorporated in many successful products such as
PARADOX, DBASE IV, ACCESS, and many more.
At Hewlett-Packard, Moshe is
currently involved in developing the IC-BY-EXAMPLE language - a new paradigm to
enable non-programmer professionals to construct their own applications.
Tutorial D
WE 13 09, 9:00 - 12:30
OQL: Language,
Compatibility with SQL, Implementation
Sophie Cluet (INRIA)
Abstract
OQL is the standard for object database query
languages proposed by the ODMG. It borrows SQL syntax but has a functional
nature that makes it simpler to use, implement, and document. Improving its
compatibility with SQL(92) and clarifying its relationship with SQL(3) will
strongly help its general acceptance by the database community. More important,
ANSI(X3H2) and ODMG are working on a proposal for a common standard involving
the current OQL.
The tutorial addresses a public of both, industrials and
researchers. Although standards are clearly of primary interest to industrials,
their evolution also provides challenging issues that should not be overlooked
by researchers.
The tutorial consists of three parts. First, we present
OQL(1.2) in details and explain how to construct simple and complex OQL queries.
Next, we discuss the OQL vs. SQL approaches and study different possibilities by
which OQL can reach full compatibility with SQL. Finally, we briefly consider
the language implementation.
About the Instructor
Sophie Cluet is currently a researcher in the Verso team at
INRIA-Rocquencourt. She holds a PhD from the University of Paris-Sud. She is one
of the designers of OQL and has implemented for O2 Technology the language's
first interpreter and optimizer. She is still very much involved in this area
(working on a future release of OQL and new OO optimization techniques).
Tutorial E
WE 13 09, 14:00 - 17:30
Transformation-based
Database Engineering
Jean-Luc Hainaut (University of Namur)
Abstract
Transformation-based software engineering has long
been considered as a major scientific approach to build reliable and efficient
programs. According to this approach, abstract specifications can be converted
into correct, compilable and efficient programs by application of selected,
correctness-preserving operators called transformations. In the database
engineering realm, transformations that are proved to preserve the correctness
of the origin specifications have been proposed in virtually all the activities
related to data engineering : schema normalization, DBMS schema development,
schema optimization, model translation, schema integration, view derivation,
proving schema equivalence, data conversion, reverse engineering, and others.
The tutorial is a contribution to the systematic study of both fundamental
and practical aspects of database schema transformations. It introduces the
fundamentals of database transformations, then it develops a practical toolset
of neutral techniques. These techniques are used to revisit the main database
engineering processes. The description of a transformation-based CASE tool and
the presentation of a reengineering case study demonstrates the applicability of
this general approach.
About the Instructor
Jean-Luc
Hainaut is a professor in Information Systems and Databases at the Institute of
Informatics of the University of Namur, Belgium. His research interests comprise
database technology, database methodology, reverse engineering, information
system evolution and CASE technology. Currently, he is heading the multinational
DB-MAIN project the purpose of which is to develop general methodologies and
CASE tools to assist practitioners in solving database application maintenance
and evolution problems.
He is the author of two books in database and DSS
design, and of over 20 recent conference papers, most of them dealing with
transformation techniques. He has presented several tutorials (including three
on database reverse engineering and reengineering) in major international
conferences on Databases and Information Systems.
Tutorial F
TH 14 09, 9:00 - 12:30
Repository Internals
Philip A. Bernstein (Microsoft Corporation)
Abstract
A
repository is a metadata manager that supports design tools for CASE, CAD and
the like. It runs as an application on a DBMS and provides three main functions
beyond those of a standard DBMS: relationships, version and configuration
management, and extensibility. We explain how these functions can be used by
design tools, and what technical requirements are implied by those usage
patterns. We then delve into the technical capabilities of each function. For
example, relationships are object-valued properties that have special update
semantics and can be scoped by configurations. Version and configuration
functions allow tools to run either on repository objects directly or on files.
Configurations support transitive containment (unlike file systems) and can
scope name spaces. A tool vendor can extend the repository's model by adding
behavior to standard repository methods on repository objects, e.g., by adding
state to the object, checking constraints on the object, and adding methods to
the objects.
The presentation will draw from examples of Microsoft's and
others' CASE tools, and where appropriate will be couched in the OLE object
model.
About the Instructor
Philip Bernstein is a
repository architect at Microsoft, where he leads a joint design effort with
Texas Instruments to design an open repository for CASE. He was formerly an
architect at Digital Equipment Corp., a professor at Harvard University and Wang
Institute of Graduate Studies, and VP Software at Sequoia Systems. Dr. Bernstein
has published over 70 articles on the theory and implementation of database
systems, is coauthor of "Concurrency Control and Recovery in Database Systems"
(Addison-Wesley, 1987). In 1994, he was awarded the ACM SIGMOD Innovations Award
for "contributions to the development and use of database systems."
Tutorial G
TH 14 09, 14:00 - 17:30
Temporal Databases
Christian S. Jensen (Aalborg University)
and
Richard T. Snodgrass (University of Arizona)
Abstract
Temporal database technology has reached a level
of maturity and sophistication that clearly illustrates that the support for
temporal data management provided by current database products is very far from
what is achievable.
The tutorial provides an overview of important issues
and concepts within temporal database management. It reviews fundamental data
modeling concepts and surveys the implementation of temporal database systems.
It also features the consensual temporal query language TSQL2 that has recently
been designed by an international committee of database researchers and
practitioners. TSQL2 adds temporal support to the current SQL standard and is
the most complete temporal query language to date.
The tutorial is relevant
for practitioners who work with databases containing time-varying data, and for
database researchers who are interested in temporal data management.
About the Instructors
Christian S. Jensen is currently
an Associate Professor at Aalborg University, Denmark. He has spent 4 years at
the University of Maryland and the University of Arizona. Recently, he
coordinated and headed two editorial boards for the ongoing consensus temporal
database glossary initiative; he is editor for the temporal database benchmark
initiative; and he was a member of the TSQL2 Language Design Committee.
Richard T. Snodgrass received his Ph.D. from Carnegie Mellon University in
1982 and joined the University of Arizona in 1989. He has written or co-edited
three books, including "Temporal Databases: Theory, Design and Implementation.'"
He chaired the TSQL2 Language Design Committee, and is now working closely with
the ANSI and ISO SQL3 committees to add temporal support to that language.
