Emerging Database Technologies
and Applications
Progression of Database Technology
1960s to Mid1970s
Network
Hierarchical
1970s to Mid-1980s
1980s to Early 1990s
Future
Relational
Database
Hardware
Mainframes
Mainframes
Minis
PCs
Merging data models with
knowledge representation
Hybrid models
Client-server configuration
Parallel processing
Optical memories
User Interface
None
Query languages
Forms
Semantic
Object-oriented
Logic
Faster PCs
Workstations
Database machines
Back ends
Graphics
Menus
Query-by-forms
Program
Interface
Procedural
Embedded query
languages
Standardized SQL
4GL
Logic programming
Presentation
and Display
Reports
Report generators
Business graphics
Image output
Processing
Processing
data
Data Model
Multimedia
Natural languages
Speech input
Freehand text
Integrated database and
programming languages
Generalized presentation
managers
Distributed, heterogeneous
data and knowledge
processing with multimedia
information
Information and
Transaction processing Parallel database
transaction processing Knowledge processing management
Current Trends in Technology
•
•
•
•
Distributed, heterogeneous environments
Open systems
More functionality
Parallel database management
Next Generation of DBMSs
• Active databases
– Applications such as process control, power
distribution/generation, workflow control, program
trading, battle management, patient monitoring are not
well served by passive DBMSs
– conditions defined on states of the database must be
monitored and actions taken
– active databases support condition monitoring
Active Databases
Specification of events and
conditions to be monitored
Queries and updates
Actions
Active DBMS
External events
Payoffs:
• No changes to applications
• DBMS optimizes rules
• Enhances DBMS functionality
• Meet the time-constrained
requirements of applications
Issues in Active Databases
• Efficiency
– a large set of rules need to be managed and evaluated
efficiently
• Modes of rule execution
– rules can be fired in an immediate, deferred, or detached
mode in regard to the original transaction
• Data model extension
– specifying events, conditions, and actions
• events:
– database operations (insert, delete, modify)
– temporal events (5 p.m. every day)
– user- or application-generated events (hardware failure)
Issues in Active Databases
• Management of rules
– ability to manipulate rules (add/delete/modify)
– mechanisms for enabling and disabling rules or rule sets
• Supporting DBMS functions
– examples: constraint management, maintenance of
derived data, rule-based inferencing
• Interaction with parts of DBMS
– optimization of rules requires interaction with transaction
manager, object manager, and scheduler
State of the Art in Active
Databases
• HiPAC (High Performance ACtive database system)
research project at Xerox
• PROBE for battle management application (Computer
Corporation of America)
• Event/Trigger Mechanism (Univ. of Karlsruhe)
• POSTGRES (Stonebraker, UC Berkeley)
• Starburst project at IBM
• Sybase supports simple triggers
• InterBase does not impose most of the restrictions seen in
Sybase
• ORACLE v. 7, INGRES, INFORMIX, etc. provide some degree
of rule and trigger support
Multimedia Databases
• Applications:
–
–
–
–
–
documents and records management
knowledge dissemination
education and training
marketing, advertising, retailing, travel
real-time control and monitoring
Multimedia Databases
• Multimedia IS are very complex; issues:
–
–
–
–
–
modeling, dealing with complex objects
design (conceptual, logical, physical) not researched yet
storage on standard devices presents problems
retrieval opens up many issues
performance problem solving efforts are experimental
• Databases (fixed data structure) versus
information retrieval (text) perspectives
• Requirements of multimedia/hypermedia data
modeling and retrieval
– query mechanism should have access to the links (?)
Multimedia Databases
• Indexing of images
– automatic object identification
– manual indexing
• Open problems in text retrieval
Multimedia information systems promise to bring
about a marriage of the disciplines of information
retrieval and database management
Spatial Database Management
• The spatial semantics can be captured by three
common representations:
– solid representation
• the space is divided into pieces
– boundary representation
• the spatial characteristics are represented by line
segments or boundaries
– abstract representation
• relationships with spatial semantics, such as ABOVE,
NEAR, IS NEXT TO, BEHIND, are used to associate entities
• The PROBE project provided support for spatial
data
Temporal Database Management
• An one-dimensional case of spatial information
• Includes three types of support for time:
– time points
– time intervals
– abstract relationships (before, after, during,
simultaneously, concurrently, ...)
• The history aspect of databases is important for
project management, patient histories,
maintenance histories, etc.
Temporal Database Management
• A range of businesses (ex. finance, medical, legal,
manufacturing) can benefit from quick access to
historical and current data
• Limitations of current databases:
– data become valid at the time they are recorded; no
provision for distinguishing between transaction time
and valid time
– no capability to preserve historical information
• Until recently, inefficient storage capabilities made
the temporal database concept not practical
• WORM and compression technology made it
possible
Tuple Time Stamping
One proposed model is the Temporal Relational
Model
– Attributes, relations are divided into time-varying and
non-time-varying
– For time-varying relations, two timestamp attributes are
appended
– SQL is extended into Temporal SQL (TSQL)
EmpNo
Salary
Position
Start time End time
33
20K
Typist
12
24
33
25K
Secretary
25
35
45
27K
Jr. Engineer
28
37
45
30K
Sr. Engineer
38
42
Open Problems
• Reasoning with temporal information
• Processing information over valid-time and
transaction-time databases
• Mixing temporal processing with active and
deductive databases
• Integrating temporal information over
heterogeneous environments
Update on Temporal DB Research
Recent Advances in Temporal Databases, J. Clifford and A. Tuzhilin,
(Eds.), Proceedings of the International Workshop on Temporal
Databases, Zurich, Switzerland, 17-18 September 1995, Springer, 1995.
X, 362 pp. 80 figs. Softcover $79.00 ISBN 3-540-19945-4
•
•
•
Papers cover a wide range of topics from the highly theoretical
through to reports on how temporal data bases can be used to solve
real problems
In addition to the technical papers, there are also summaries of two
panel discussions which assess the recently-completed TSQL2
Language Design, and examine the need for additional research into
the development of TSQL3
Together these papers provide a comprehensive overview of the latest
research work into the area of temporal databases
Also see URL: http://www.jcc.com/sql_tmpr.html
Extensible Database Management
• Building DBMSs out of “DBMS parts”
• Assembling prewritten modules has advantages:
– rapid and economical development
– technological improvements can quickly be
incorporated
– proposed new algorithms can first be evaluated
• Project GENESIS at the University of Texas
– components of DBMS and interfaces among them are
defined
– new DBMS can be configured within minutes
– plug-compatible modules are defined for access
methods, query optimization, concurrency control,
recovery, ...
Extensible Database Management
Project EXODUS at University of Wisconsin provides certain kernel
facilities including storage manager and type manager
– type manager permits definition of hierarchies with multiple
inheritance
– the storage object is a byte sequence of arbitrary size
– buffer management, concurrency control, recovery mechanisms
are provided and can be modified
– type-independent index structures can be selected
– the language E, an C extension, adds the notion of persistent
objects
– query processing includes query optimization and evaluation
– the DBI supplies the description of operators and methods to
implement them
– the rule-based optimizer generates C source code
– the EXODUS storage manager was used by several vendors (incl.
O2 )
Full-Functionality Approach
• Building DBMS with extensive functionality
• Providing a wide set of features
• Projects PROBE and Starburst
– active databases
– PROBE provides spatial query processing
• POSTGRES also combines OO and active
database capabilities with the relational model
Unified Database Management
Example of systems in this category is UniSQL
– combines
• power and ease of popular development tools
• OO development
• multimedia database integration
– organization:
•
•
•
•
UniSQL/X provides C/S DBMS platform
UniSQL/M allows access to relational and prerelational DBs
UniSQL/4GE Tools for dynamically generating applications
Visual Editor and Media Master allow for viewing and editing
of schemas and for sophisticated report generation
– the next generation of DBMSs is likely to be patterned
after the UniSQL