Course Notes on

Databases and Database Management Systems

Databases and Database Management Systems: Summary
Databases
Database management systems
Schema and instances
General view of DBMS architecture
Various levels of schema
Integrity constraint management
Notion of data model
Database languages and interfaces
Other DBMS functions
Roles and functions in database management
1
Database-Management Systems, October 7, 2008 1
Database
A collection of related data with
3 logically coherent structure
3 inherent meaning
3 purpose, for intended users and applications
3 varying size
3 scope, content of varying breadth
3 physical organization of varying complexity
3 various applications with possibly-conicting objectives
3 persistence, existence over a long period of time
2
Database = a collection of related data with
3 a logically coherent structure (can be characterized as a whole)
3 some inherent meaning (represents some partial view of a portion of the real
world)
3 a specic purpose, an intended group of users and applications (a database
embodies a biased, operational view on the world; database management is
not after modeling the world in general, maybe philosophy or ontology are)
3 a largely varying size (from a personal list of addresses to the National Register
of Persons)
3 a scope or content of varying breadth (from a personal list of addresses to a
multimedia encyclopedia)
3 a physical organization of varying complexity (from a manual personal
list, managed with simple les, to huge multi-user databases with geographically
distributed data and access)
3 logically-coordinated objectives, data is dened once for a community of users,
and accessed by various applications with specic needs
Database-Management Systems, October 7, 2008 2
Database Management Systems (DBMSs)
DBMS: a collection of general-purpose, application-independent programs
providing services to
3 dene the structure of a database, i.e., data types and constraints that
the data will have to satisfy
3 manage the storage of data, safely for long periods of time, on some storage
medium controled by the DBMS
3 manipulate a database, with ecient user interfaces to query the database
to retrieve specic data, update the database to reect changes in the world,
generate reports from the data
3 manage database usage: users with their access rights, performance op-
timization, sharing of data among several users, security from accidents or
unauthorized use
3 monitor and analyze database usage
3
DBMS have similarities with operating systems: both manage memory, process schedul-
ing, I/O, communication
In addition, DBMSs implement many data-management functions
Other name for DBMS: database system, database manager
DBMSs typically do not use the le system of the operating system of the machine
where they are installed. Instead, the dene their own richer le organizations and
access methods
Database-Management Systems, October 7, 2008 3
Example of a Database
Student
StudName StudNo Class Dept
Smith 17 1 CS
Brown 8 2 CS
Course
CourseName CourseNo Credits Dept
Introduction to CS CS1310 4 CS
Data Structures CS3320 4 CS
Discrete Mathematics MA2410 3 MA
Database Management CS3380 3 CS
Prerequisite
CourseNo PrereqNo
CS3380 CS3320
CS3380 MA2410
CS3320 CS1310
Session
SessIdent CourseNo Semester Year Professor
85 MA2410 Fall 96 King
92 CS1310 Fall 96 Anderson
102 CS3320 Spring 97 Knuth
112 MA2410 Fall 97 Chang
119 CS1310 Fall 97 Anderson
GradeReport
StudNo SessIdent Grade
17 112 14
17 119 12
8 85 16
8 92 16
8 102 14
4
Important Functions on a Database
Structure denition: declare 5 les or relations + data types, e.g.
Student(StudName, StudentNo, Class, Dept)
Population: input data about specic students, courses, prerequisites, . . .
Querying
3 Which are the prerequisites of the Database course ?
3 List students who got grade 14 or 16 for the Database course in 1993
Reporting: prepare diplomas, with standard text, interspersed with name of
student, courses taken, name of degree, grades, etc.
Modication, update of population
3 Create a new session for the Database course
3 Enter a grade 16 for Smith in the Database Session
Modication of structure, of schema
3 Create a new relation for instructors
3 Add Address attribute to relation Student
5
Database-Management Systems, October 7, 2008 4
Transient and Persistent Data
In practice, information systems often require persistent data
Data: relevant facts about the domain of interest
3 persistent: continues to exist even when the system is not active
3 transient: created while an application is running and not needed when the
application has terminated
Persistent data must be stored in secondary memory (not just in computer mem-
ory) and organized to be made available to several applications
6
Data and Database Schema
Fundamental hypothesis of database modeling: the information contained
in a database is represented on two levels: (1) data (large, frequently modied)
and (2) structure of data (small, stable in time)
Database schema: description of DB structure, accessible by programs
Database =

Data Type
Metadata
Structure
Schema
Intension
Catalog
Directory
Data dictionary

Instances
Occurrences
Data
Extension
Population

DBMS software is application-independent it consults the database

structure in the data dictionary to understand and execute application programs
7
Database-Management Systems, October 7, 2008 5
Ontology is another more recent term for designating the structure of an application
domain (= schema information valid for several related applications)
Various Levels of Schema
A DBMS provides users with
3 a conceptual representation of information from the point of view of users
3 a physical or internal representation with the implementation details
Both are necessary, but each has its purpose
Users refer to the conceptual representation and the DBMS ensures the corre-
spondence with the physical representation
The success of relational technology has demonstrated that physical concepts
can be hidden from users and that there are substantial advantages for doing so
8
Technical vocabulary: logical versus physical concepts:
3 logical (old terminology) or conceptual (current terminology) information: deals
with the user view on data (in terms of concepts familiar to actors in the appli-
cation domain)
3 physical or internal concepts: concern the implementation of conceptual con-
cepts into the hardware/software infrastructure (this is a technological area)
Database-Management Systems, October 7, 2008 6
Gross Architecture of DB and DBMS Software
Metadata
Software
DBMS
Software to Process
Queries / Programs
Software to Access
Stored Data
System
Database
Application Programs /Queries
Users / Programmers
Data
9
Data Structures in the Conceptual Schema
Student
StudName StudNo Class Dept
Course
CourseName CourseNo Credits Dept
Prerequisite
CourseNo PrereqNo
Session
SessIdent CourseNo Semester Year Professor
GradeReport
StudNo SessIdent Grade
10
Database-Management Systems, October 7, 2008 7
In addition to data structures, the schema also comprises
3 the denition of domains for data elements (attributes)
3 the specication of constraints, to rene the data-structure part of the schema
Data Structures in the Physical Schema
Student records are stored in a le as follows:
Data Item Name Starting position Length (bytes)
Name 1 30
StudentNumber 31 4
Class 35 4
Department 39 4
3 length of Student records = 42 bytes
3 the le is ordered by values of the Name eld
3 the le is indexed on Name
11
Database-Management Systems, October 7, 2008 8
Support of External Views
A single database usually serves the needs of a community of users dierent
perspectives or views on the same data are often natural
View: some subset of the database or some restructuring of the database suited
for an application
Views are redundant with the basic denition of the database (they may or may
not be stored explicitly, this is an eciency issue)
DBMS takes care of the correspondence between views and database data
View denitions are part of the database schema
Other terms for view: subschema, external schema, derived relation
12
More on Views
Relations in a database comprise base relations and views (users do not necessarily
have to know which is which)
Applications access the database through views, without necessarily knowing about
the whole database
Views may be
3 queried
3 combined in queries with base relations
3 used to dene other views
3 in general, NOT updated freely
Database-Management Systems, October 7, 2008 9
A View for Preparing Diplomas
Diploma Data
StudName
StudentTranscript

CourseName Grade Semester Year

Smith
Introduction to CS 12 Fall 97
Discrete Mathematics 14 Fall 97
Brown
Discrete Mathematics 16 Fall 96
Introduction to CS 16 Fall 96
Data Structures 14 Spring 97
Database Management 16 Fall 97
13
View Diploma Data is read-only (data cannot be entered into the database or modied
or deleted thru Diploma Data)
View denition, dene view (new attr.) as SELECT ... (show SQL query)
Display cross reference of views with base tables
Database-Management Systems, October 7, 2008 10
Data Approach to DBMS Architecture: ANSI 3-Schema
Architecture
Community
level
Internal
level
Internal Schema
Stored Database
mapping
Community Schema
External View 1
Community / internal mapping
External View n
End users
level
External
External / community
14
ANSI 3-Schema Architecture = general DBMS architecture to maintain several de-
scriptions of the data in a database
3 levels of schemas
3 external schemas (user views) dene the relevant data for application programs
and hide the rest of the database
3 the community schema describes the common conceptual structure of the
whole database; it contains and integrates the information contained in all the
user views
3 the internal schema describes the database storage and access structures
3 data actually exists only at the internal level, it is accessed from the external
level; DBMS provides mappings (compiled or interpreted) between levels, in
both directions
Note on vocabulary
3 in 1975, ANSI (American National Standards Institute) called conceptual sche-
ma what we call community schema here
3 nowadays, conceptual is more often used in another sense: a conceptual
schema describes the user view of information, independently of the data model
of the DBMS and its implementation
3 the conceptual schema is produced by the analysis phase in the process of database
design
Database-Management Systems, October 7, 2008 11
3-Schema Architecture and Data Independence
Data independence: possibility to change the schema at one level without
having to change it at the next higher level (nor having to change programs that
access it at that higher level)
3 logical data independence: an external schema (and programs that access
it) is insulated from changes that does not concern it in in the community
schema (and in the physical schema)
3 physical data independence: the community and external schemas are
insulated from changes in the physical schema
15
With le processing, changes to le structure entail changes to application programs
(e.g., COBOL mixes all three levels in its data division)
Data independence did not come easy (one of the great debates of the relational
revolution was about whether data independence could be realized with reasonable
eciency)
Program/data independence is sometimes used instead of data independence, to
emphasize that application programs remain unchanged when some changes are made
to the data
Database-Management Systems, October 7, 2008 12
Why Data Independence is Important
or Some Virtues of Abstraction
It is a divide-and-conquer strategy to help master complexity and think
precisely (user programs are more abstract, higher-level, simpler, and shorter)
It leaves open more possibilities for the system to optimize implementation
strategies
It contributes to stability in time for applications (fewer changes in data struc-
tures to adapt to)
16
Pre-relational (pre-historical) information systems went thru monolithic applications,
combining in the same program user-interface management, implementation of busi-
ness logic, data processing, and persistent-storage management
The lack of data independence caused very high human cost in terms of re-programming,
especially for evolution and maintenance, i.e., modications to operational information
systems for adapting to changes in requirements and changes in the system environ-
ment
Database-Management Systems, October 7, 2008 13
Examples of Data (In)dependence
Two elds within the same record: conceptual (relevant semantic link) or phys-
ical (clustering for fast joint access)
A link between two records: conceptual (relevant semantic link) or physical (a
pointer for fast navigational access)
Ordering output data: how this is implemented (physical ordering or sorting) is
left to the DBMS and invisible in application programs
Adding a physical index to speed up an application should not require modifying
the application program, the only visible eect will be eciency
Add a new eld to a le: only programs that access the new information need
be modied
17
Constraints
Integrity constraint: any prescription (or assertion) on the schema (i.e., valid
for all extensions of the database, now, in the past, and in the future)
Constraints model extra information not denable in the data-structure part of
the schema
Constraints cannot be deduced from the database extension, they can only be
veried, checked
Examples
3 data types: grades are integers between 0 and 20
3 uniqueness of values: no two students have the same student number
3 referential integrity: all StudNo values in GradeReport must also appear as
values of StudNo in Student
18
Database-Management Systems, October 7, 2008 14
Consistency constraint is a better term than integrity constraint
Any piece of information can be given the status of constraint (i.e., belong to the
schema); remember that
3 schema information is normally more stable in time than instance data
3 schema semantics is managed by DBMS software (still, not all constraints are
managed by DBMS software, see later)
Constraints
Some constraints express a part of the general semantics of the data, i.e.,
of the application domain modeled in the database
Some constraints are technical (e.g., referential integrity is specic to the rela-
tional model)
Not everything can be controled by constraints (e.g., misspellings in names can
only be checked manually)
Data models with richer, more expressive data structures have fewer explicit
constraints than simpler ones (see later)
Constraints have to be checked when updates are performed on the database
19
Basic rule of database modeling:
3 the data in the database must conform to both the prescriptions of the database
structure (the schema) and the constraints
3 it is often said that the constraints belong to the schema
3 basic modality: every piece of data that does not contradict the prescriptions of
schema + constraints is acceptable in the extension of the database (everything
that is not explicitly forbidden is permitted)
Database-Management Systems, October 7, 2008 15
Constraints are Properties of Data, not of Applications
Semantics
Application
Semantics
Application Application Application
Data
Data and
Semantics
File-based approach Database approach
20
A progressive evolution of DBMS technology from traditional le processing is to move
constraints as much as possible from application programs into the schema
Database-Management Systems, October 7, 2008 16
Constraint Enforcement in Application Programs
(DBMS)
Processing
Update
else ...
then insert new Session
if exists
Check Course Number
User Program
Schema Data
21
Example: application program that adds a new Session for a Course in the presence
of a referential constraint, that requires that the CourseNo must exist
The program is given as input a 5-tuple of values: (SessIdent, CourseNo, Semester,
Year, Professor)
Database-Management Systems, October 7, 2008 17
Constraint Enforcement by DBMS
Schema
+
Constraints
Data
else ...
then insert new Session
if exists
Check Course Number
(DBMS)
Processing
Update
IC
Processing
(DBMS)
insert new Session
User Program
DBMS enforcement of constraints simplication of user programs at the cost
of more complexity for DBMS software
22
Various Semantics in Modeling
(1) Schema semantics: used and maintained by system software (data structures,
consistency constraints)
(2) Instance semantics: informal, intuitive real-world semantics; not used nor
maintained by system software
(3) Denotation semantics: relates elements in (1) with corresponding elements
in the real world; makes explicit the relationship between (1) and (2)
(4) Environment semantics: relates elements in (1) and their corresponding real-
world elements with system environment in the real world
(3) and (4) are usually informal documentation in the data dictionary
23
Database-Management Systems, October 7, 2008 18
In the database approach, the schema expresses the general semantics of the part
of the world that is modeled (in terms of data structures, concepts, categories, con-
straints) while the data expresses the semantics of individual objects
The instance semantics is not modeled nor exploited by the tools (i.e., the DBMS):
it is left to the interpretation of users, possibly helped by programs (e.g, to consult an
informal denition)
Another part of the semantics deals with the correspondence between schema
and real world. It is rarely exploited by tools and is at best represented informally
in design documents and in the data dictionary of the database (documentation)
General semantics (in the schema) comprises
3 the data structures for the application domain in the data model of the DBMS
3 classical constraints expressing regularities in the application domain (e.g.,
uniqueness, keys, referential integrity, normal forms)
3 ad-hoc constraints (e.g., there are no two dierent sessions of the same course
with the same professor, or, all CS students take at least 3 CS courses)
Extracting the relevant general semantics of an application domain is the process of
database design; it results in a conceptual schema which is the basis of an agreement
between domain analysts and users, as well as the starting point for implementing the
database
Data Model
Data model = languages for dening structure and behavior
3 structure: data types, relationships, constraints
3 behavior: basic operations for retrievals and updates
DBMSs support several levels of data models:
3 conceptual: closer to user view on data (nowadays, mostly entity-relationship)
3 physical: for storage structures and access methods
3 logical or implementation: (historical) compromise between conceptual
and physical organization
Object-oriented models allow to integrate user-dened structure and behav-
ior (but object orientation in the database world is complicated)
24
The relational model was the rst model to be dened: the concept of data model is
an important contribution of the relational era
Database-Management Systems, October 7, 2008 19
Although it clearly distinguishes between information content from the user point of
view and its implementation, the relational model cannot be considered as a concep-
tual model
More semantic data models are appropriate as conceptual models for database design
(currently, typically, the entity-relationship model)
The relational model has become a compromise: logical relational schemas (e.g., in
SQL) give a simple view of data as relations (or tables, or abstract les)
The implementation models of commercial DBMSs are mostly relational, although
network and hierarchical systems are still used
Network and hierarchical models were dened after the fact, as more or less satis-
factory a posteriori abstractions
Data Models and their Implementation
Data model:
3 abstract, self-contained mechanisms for dening data structures and opera-
tions
3 hides low-level storage details abstract machine for user interaction
Implementation of the data model: physical realization on a computer archi-
tecture
The distinction relates to physical data independence (distinction between logical
and physical concepts)
25
Database-Management Systems, October 7, 2008 20
Database Languages
Data Denition Language (DDL), for
3 writing all schemas and mappings between schemas
3 specifying constraints
Data Manipulation Language (DML), for
3 database manipulation (retrieval, insertion, deletion, modication) with query
languages and programming languages
3 user-friendly interfaces (graphical, menu-based, forms-based, natural lan-
guage, parametric)
26
Levels of DML Languages
High level (e.g., SQL)
3 specify what data is to be retrieved rather than how to retrieve it
3 used on their own or embedded (data sublanguage) in a programming
language (host language) like C, Pascal, COBOL
3 also called declarative, assertional, nonprocedural, set-at-a-time, set-oriented
Low level
3 retrieve individual records and process each one separately
3 also called procedural, or record-at-a-time, navigational
27
Database-Management Systems, October 7, 2008 21
Database Functions and Application Functions
Database functions or DBMS functions: supplied by the DBMS and invoked
in application programs
Application-program functions: to be programmed in application programs
Evolution:
3 the power of DBMS software is continuously increasing
3 more and more functions that used to have to be programmed are progres-
sively turned into DBMS functions, because
some data-management issues are better understood and can be turned
into DBMS modules
the processing ressources continuously increase
28
Other DBMS Functions
Concurrency control
Backup and recovery
Redundancy management
Access control
Performance optimization
Metadata management
Active features (rules, triggers)
29
Database-Management Systems, October 7, 2008 22
DBMS Function: Concurrency Control
Transaction processing (OLTP) applications (e.g., banking and airline sys-
tems), with multiple simultaneous users
Concurrency control: ensures correctness of competing accesses to same data
Correctness: 4 desirable properties (A.C.I.D.)
3 Atomicity: all or nothing, transactions execute entirely or not at all
3 Consistency: transactions move the DB from a consistent state to a consis-
tent state
3 Independence or isolation: no partial eects of incomplete transactions
are visible
3 Durability: successfully-completed transactions are permanent, cannot be
undone
30
Transaction = one execution of a user program (executing the same programs several
times corresponds to several transactions)
Transaction = basic unit of change as seen by the DBMS
3 partial transactions are not allowed (atomicity)
3 the eect of a collection of transactions is equivalent to some serial execution of
the transactions (serializability)
Database-Management Systems, October 7, 2008 23
Three Transactions
T1 Read NP
NP NP-1
Write NP
T2 Read NP
NP NP-1
Write NP
Read NQ
NQ NQ+1
Write NQ
T3 Read NP
...
Read NQ
...
Read NP
31
T1 could be the reservation of an instance of a resource, e.g., a seat for a particular
ight; NP is then the number of seats available on that ight; T2 is the canceling of
a reservation (NQ) combined with a reservation (NP); T3 just queries the database
Read NP is a transfer of information from the database to the user space
NP NP - 1 is performed in the user space (i.e., with no eect on the database)
Write NP modies the database
Database-Management Systems, October 7, 2008 24
An Incorrect Schedule
Step T1 T2 T3
1 Read NP
2 Read NP
3 NP NP-1 Read NP
4 Write NP
5 Read NQ
6 NP NP-1
7 Write NP Read NQ
8 NQ NQ+1
9 Write NQ Read NP
32
Both T1 and T2 work with the same value from the database
The update by T2 is not preserved in the database
Database-Management Systems, October 7, 2008 25
A Correct Schedule
Step T1 T2 T3
1 Read NP
2 NP NP-1 Read NP
3 Write NP
4 Read NP
5 NP NP-1
6 Write NP Read NQ
7 Read NQ
8 NQ NQ+1
9 Write NQ Read NP
33
Correctness is obtained by sequencing potentially conicting updates
Database-Management Systems, October 7, 2008 26
Which Level of Concurrency Control?
Tradeo between
3 eciency (# transactions/sec), and
3 cost of maintaining consistency
Complex, powerful piece of engineering (particularly for geographically distributed
and physically redundant databases) coupled with ne management optimization
Example: how much do we want to protect against
3 two simultaneous withdrawals from the same bank account
3 multiple reservations of the same airplane seat
34
DBMS Functions: Backup and Recovery
DBMSs provide facilities for recovering from hardware and software failures
If the computer system fails during a complex update program
3 the DB must be restored to its state before the program started, or
3 the program must be resumed where it was interrupted so that its full eect
is recorded in the database
More complex and important in a multi-user environment
35
Database-Management Systems, October 7, 2008 27
DBMS Function: Redundancy or Replication Management
Redundancy: storing several copies of the same data
Frequent in traditional le processing: a goal of the database approach was to
control redundancy as much as possible
Problems with redundancy
3 waste of storage space
3 duplication of eort to perform a single conceptual update
3 danger of introducing inconsistency if multiple updates are not coordinated
Replication of the same data may be useful for optimizing physical accesses
(typically in distributed databases)
36
Controling redundancy was a major progress of database technology over le systems
If the right decisions can be made during database design, the resulting central
database schema presents a more uniform data representation (e.g., avoids dierent
types for the same data as typically happens with le systems)
If present for eciency, redundancy should remain invisible to ordinary users and be
under the control of the DBMS (a complex technical problem in general)
Database-Management Systems, October 7, 2008 28
DBMS Function: Access Control
Who accesses what data, to do what, when, from where, etc.
Access control is mandatory in a multiuser database, e.g., for condentiality
Various access modes to data (e.g., read only, read and update)
DBMS subsystem enforces security and authorization
Restrictions concern programs (e.g., who can create new bank accounts) and
data (e.g., which bank accounts can I see)
37
The data dictionary holds information about users and their access privileges (e.g.,
name and password)
Several levels of access privileges
3 to create a database
3 to authorize (grant) additional users to
access the database
access some relations
create new relations
update the database
3 to revoke privileges
Database-Management Systems, October 7, 2008 29
DBMS Function: Performance Optimization
Good manual optimization of DB programming is scarce and expensive
Performance optimization is largely a DBMS function
This is made possible by
3 physical data independence
3 high-level data models with user programs that can be optimized by DBMS
software (unlike navigational record-at-a-time programs for which optimiza-
tion can only be manual, i.e., left to users)
DBMS maintains information (metadata) on database populations, in addition
to storage structure (conceptual schema) and access paths (physical schema)
Actual optimum varies with evolution of DB population: physical reorganiza-
tions are sometimes necessary (e.g., add index, drop index, sort le)
38
Ecient relational query optimization was a key to the acceptance of the relational
model in the 80s
What matters today is human performance, not machine performance
Database-Management Systems, October 7, 2008 30
DBMS Function: Metadata Management
Data about data is also data: metadata
System catalog (or data dictionary): special DB maintained by DBMS
Information in the catalog: data objects, DB statistics, physical structures and
access paths, user access privileges, etc.
Accessible to DBMS functions and to users
39
Example Catalog of Relational Schema
RelationAttributesCatalog
RelName AttrName AttrType FKMember FKRelation
Employee SSN String(9) no
Employee FName String(15) no
Employee MInit char no
. . . . . . . . . . . . . . .
Department DName String(10) no
Department DNumber Integer no
Department MgrSSN String(9) yes Employee
. . . . . . . . . . . . . . .
RelationKeys
RelName KeyNumber MemberAttr
Employee 1 SSN
Department 1 DNumber
Department 2 DName
. . . . . . . . .
40
Database-Management Systems, October 7, 2008 31
Example Catalog of Relational Schema
RelationIndexes
RelName IndexName MemberAttr IndexType AttrNumber AscDesc
WorksOn ESSNIndex ESSN clustering 1 Asc
WorksOn EPIndex ESSN secondary 1 Asc
WorksOn EPNIndex PNO secondary 2 Asc
. . . . . . . . . . . . . . . . . .
ViewQueries
ViewName Query
OldEmps Select SSN, Fname,LName
From Employee
Where BDate < 01/01/1950
. . . . . .
ViewAttributes
ViewName AttrName AttrNumber
OldEmps SSN 1
OldEmps FName 2
OldEmps LName 3
. . . . . . . . .
41
Active-Database Technology
Passive DBMS: all actions on data result from explicit invocation in applica-
tion programs (they only do what application programs tell them to do)
Active DBMS: execution of actions can be automatically triggered in response
to monitored events, including
3 database updates: upon deletion of the data about a customer
3 points in time: on January 1, every hour
3 events external to the database: whenever paper jams in the printer
42
Database-Management Systems, October 7, 2008 32
Evolution of database technology has been going thru representing and supporting
more functionality of database applications within the DBMS, e.g.,
3 checks of some types of integrity constraints (produced from a declarative
denition located with the database schema)
3 stored procedures: precompiled procedures located within the database, in-
voked from application and system programs
3 common semantics abstracted from application domains (e.g., for spatial, mul-
timedia, temporal, deductive, active databases)
Active-database technology
3 a relatively recent extension of traditional DBMS technology
3 most commercial RDBMSs include some capability for rules or triggers
3 research prototypes provide more comprehensive support for active rules than
RDBMSs
Application semantics in programs for active DBMSs is expressed in:
3 traditional application programs (as for passive DBMSs)
3 rules (in the database, available to all applications)
Event - Condition - Action Rules
When an event occurs, if a condition holds, then an action is performed
Event
a customer has not paid 3 invoices at the due date
Condition
if the credit limit of the customer is less than 20 000 Euros
Action
cancel all current orders of the customer
ECA rules are part of the database (rule base), available to all applications
43
Database-Management Systems, October 7, 2008 33
Rules May Express Various Aspects of Application Semantics
Static constraints (e.g., referential integrity, cardinality, value restrictions)
3 only regular students can register at the library
3 students can register in no more than 20 courses
3 the salary of employees cannot exceed the salary of their manager
Control, business rules, workow management
3 when data for new students is recorded, data is automatically entered to
register the students in the mandatory courses
3 all expenses exceeding 50K must be approved by a manager
3 when an order has been accepted, an invoice is sent
Historical data
3 the data about completed orders is transferred monthly to the data warehouse
44
Database-Management Systems, October 7, 2008 34
Semantics Modeled by Rules (contd)
Implementation of generic relationships (e.g., generalization)
3 a person is a student or a lecturer, but not both
Derived data: materialized attributes, materialized views, replicated data
3 the number of students registered in a course must be part of the course data
3 orders received are summarized daily in the planning database
Access control
3 employees can view data about their own department only
Monitoring: performance, resource-usage monotoring
3 the number of disk accesses of each database query is recorded and statistics
are produced weekly
3 each access to our web pages is reected in the usage database
45
Exercise: rephrase the above examples as event-condition-action rules
Note that many examples have a more declarative form than ECA rules
Database-Management Systems, October 7, 2008 35
Benets of Active Technology
Simplication of application programs: part of the functionality can be
programmed with rules that belong to the database
Increased automation: actions are triggered without direct user intervention
Higher reliability of data thru more elaborate checks and repair actions
better computer-aided decisions for operational management
Increased exibility thru centralisation and code reuse reduced develop-
ment and maintenance costs
46
People around DBs/DBMSs
Casual users
Parametric users
Application programmers
Database designers
Database administrator (DBA)
DBMS vendors
System programmers
Operators
47
Database-Management Systems, October 7, 2008 36
People and Functions around DBs/DBMSs
End users:
3 casual users: occasional unanticipated access to DB (e.g., tourists, managers)
3 parametric users: query and update the database through xed programs
(invoked by non-programmer users) (e.g., banking)
Application programmers: implement database application programs that facili-
tate data access for end users
Database designers:
3 prepare external schemas for applications
3 identify and integrate user needs into a conceptual (or community, or enterprise)
schema
Database administrator (DBA):
3 dene the internal schema, dening subschemas (with database designers), and
specify mappings between schemas
3 coordinate, supervise, and monitor database usage
3 supervise DBMS functions (e.g., access control, performance optimization, backup
and recovery policies, conict management)
DBMS vendors and their technical sta (build and maintain the DBMS software)
System programmers: interact with DBMS software and internal database level
Operators: responsible for running and maintaining the HW/SW for the DBMS,
backup, recovery from failures, etc.
Points of View on DBMS Architecture
Data: several data views and their relationships (ANSI 3-schema architecture)
Components: DBMS software viewed as a number of components providing
functionality; emphasis on DBMS system design and implementation
Functions: dierent classes of users and functions performed for them by DBMS
software; no emphasis on how functions are realized
Operation: how DBMS functions are realized with current software, hardware,
and network infrastructure (client-server architecture)
48
Database-Management Systems, October 7, 2008 37
Main DBMS Components
Processor
Run-time Database
Program
Compiled
Stored Database
Concurrency control
Backup-Recovery Subsystem
Compiler
Host Language
Host language
Program
Privileged
commands
Statements
DDL
Query
High-level
Query
Compiled
Transactions
Compiled
Processor
Query
Compiler
DML
programs
Application
Compiler
DDL
Statement
DML
Catalog Data/
Dictionary
Precompiler
System
DBA Staff Casual users Application programmer Parametric users
49
DDL compiler: builds inter-schema mappings
Application program = mixture of regular program and DML statements (SQL queries)
High-level queries DML statements
Compiled program compiled transaction
Not shown: answers to queries and programs, access control, ...
Database-Management Systems, October 7, 2008 38
Advantages of the Database Approach
Summary
3 separate DBMS functions from application functions
3 move application domain semantics out of programs into DB schema
Reduced application-development time
3 simpler programs because many functions can be invoked from the DBMS
Uniformization of organizational procedures
3 at the expense of more eort on initial database design
Reduction of redundancies in personnel and procedures
3 e.g., to manage data redundancy or complex conversion processes
Centralized management of information, performance, conict resolution
Rationalization of information processing
3 users can concentrate on using information
BUT DBMS software is complex and expensive, not all applications need all
DBMS functions
50
Database-Management Systems, October 7, 2008 39