IT 20303 The Relational DBMS. Relational Database Theory Building a Database –Begin by developing a conceptual model of the application domain Based on.

IT 20303

• The Relational DBMS

Relational Database Theory

• Building a Database– Begin by developing a conceptual model

of the application domain• Based on requirements document• Described in terms of

–Entities–Data items–Relationships between entities


– Example: a mail-order catalog database

SUPPLIER PRODUCTsupplies


– Build the database, using the conceptual model as a blueprint

• Information about entities, data items, and relationships stored in data dictionary


• A database includes

– Data Tables

– Data Dictionary Tables

• Creating a table defines the table structure & characteristics in the data dictionary


• Data Dictionary

– A collection of system tables, sometimes called the catalog

– Describes the data structures, data items, & constraints


• Data Dictionary– Tightly integrated with the RDBMS

• Data definition commands create, drop, & alter data structures

–Update the dictionary definitions• Data manipulation commands query &

update the data–Access the dictionary first, then the

data


• Example of a dictionary table:

TABLE_NAME CREATER ***

Supplier User1 ***

Product User1 ***

Customer user2 ***

*** *** ***

USER_TABLES


• Evolving a Database

– Change characteristics of the database as real world, requirements change

• Without impacting the application(s)


– Requirements do Change• Add new entities, new data items• Change format, size of a data item• Change relationships

– Tune the database to address performance concerns• Add or drop an index• Change the storage allocation on disk


• Manipulating Data

– Data manipulation verbs

• Operate on a set of rows at a time

• Based on the relational algebra

• Support querying and updating data


• Manipulating Data

– Manipulating data includes the CRUD actions on data

• Create a new supplier: Insert one (or more) rows in the table

• Read data about a supplier: Select data from the table


• Update data about a supplier: Update data in one or more fields

• Delete a supplier: Delete one (or more) rows from the table

Create Insert

Read Select

Update Update

Delete Delete


• SQL: A Standard Database Access Language– SQL is a command-driven database

access language• Can be used interactively or

programmatically– SQL is a nonprocedural language

• State what you want rather than how to get it


– ANSI 1992, 1999, & ISO Standards• Describe concepts, verbs, & actions

taken for each verb

Data Definition – DBAs Data Manipulation – User Data Control - DBAs

CREATE SELECT GRANT

DROP INSERT REVOKE

ALTER UPDATE

DELETE

Database objects: tables, indexes, views, constraints, tablespaces

Data in tables Permissions on tables & users


• Concurrency & Security– Support access to the database by

multiple users simultaneously• Concurrency control is usually

implemented using locks– Provide data security by restricting

access to authorized users• Control who can access the data• Control type of access: SELECT,

INSERT, UPDATE, DELETE


• Data Integrity

– Data model reflects some part of the real world of the enterprise

– Integrity means the data accurately represents the enterprise


• Data Integrity

– Support for integrity requires support for rules about data

• Business rules about the format, use, & meaning of the data

• Action to be taken when rules are violated


• Data Integrity– Rules can also be called Semantic

Integrity Constraints– An RDBMS enforces user-defined

constraints• We define business rules to RDBMS

as constraints• RDBMS prohibits changes to data that

violate a constraint


• Integrity Constraints: Example

– Constraint involving one attribute• Salary must be a positive number

– Check (sal >0)• Dob and startdate must be valid dates

– Constraint involving two attributes in a table• Startdate cannot be less than birthdate

– Check (startdate > dob)

Empno Ename Dob Startdate Job Mgr Sal deptno


• Integrity Constraints: Example

– Constraint involving two or more rows in the same table• Department number of employee must

be the same as the department number of the manager - Trigger

• Salary of employee must be less than salary of manager - Trigger

Empno Ename Dob Startdate Job Mgr Sal deptno

Ended 09-07-05

• End Section


• An RDBMS supports two types of Constraints– Declarative Constraints – in ANSI SQL

• Stored in the data dictionary• Inherent in any fully relational RDBMS

– Defined when a table is created or altered» Not Null» Primary Key» Foreign Key, w/On Delete option» Check, for simple domain constraints


• An RDBMS supports two types of Constraints– Enterprise-Specific Rules

• Implemented as database programs• Triggers & Stored Procedures (reusable

code)


• Triggers & Stored Procedures– A Stored Procedure is a named module of

code• Stored in the database• Registered in the dictionary• Contains SQL as well as program logic• Can be invoked by any program that

can issue a request to the RDBMS, including another stored procedure or trigger


• Triggers & Stored Procedures– A Trigger is a special kind of stored procedure

• Defined on one table• Fires when there is a specified SQL request

– INSERT, UPDATE, and/or DELETE• Used to implement rules requiring logic

– Example: if qty_on_hand is less than reorder_level, call a procedure to order more products


• A Database includes Dictionary Tables, Data Tables, & Stored Programs– The RDBMS can enforce the rules of the

enterprise• Thereby protecting the integrity of the

data


• An RDBMS supports Data Recovery– Failure occurs while program is executing

• After updating some data items• Before updating related data items

– Solution is Transaction Processing


• Transaction Processing– A transaction is defined to be a logical unit

of work• Contains several accesses to the

database• Transforms database from one

consistent state to another


• Transaction Processing– Purpose of transaction processing is

atomicity of transaction• Ensures transaction executes in its

entirety or• Leave database as if nothing had ever

happened– The DBMS manages transactions and

keeps a log of all activities


• Transaction Processing– User program signals end of transaction

• COMMIT signals successful end of transaction– Makes all changes permanent– Releases all locks

• ROLLBACK signals unsuccessful end of transactions

– Discards all changes since last commit– Releases all locks


• Recovery– Recovery from program aborts

• Do Not use this if possible• Program detects error & issues

ROLLBACK• DBMS erases changes back to last

COMMIT


• Recovery– Recovery from software or hardware

failure• When DBMS is initiated, it examines

transaction log• DBMS effects ROLLBACK for all

incomplete transactions


• Recovery– Recovery from media failure (disk

information destroyed)• Database must be restored from backup

copy• DBMS must reprocess transactions on

log file since the backup (ROLLFORWARD)


• Performance– DBMS software is complex & consumes

resources• Disk input/output, the amount of data

transferred between the disk and memory– Performance measured by

• Number of transactions per minute (second)• Time (CPU, elapsed) required to perform a

function• Number of users supported, response time

experienced by users


• Performance– Factors that impact performance

• Path length, the number of CPU instructions executed


• Performance-Oriented Features– Provide efficient storage & access techniques

• Storage & access techniques are usually proprietary

– Based on indexing– Provide query optimization

• Translate nonprocedural SQL request into an optimized query plan

• Using computer science optimization techniques for relational algebra

• Using statistics in the data dictionary about the tables & indexes


• Program/Data Structure Independence– Ability to change data structures without

impacting applications– Accomplished by hiding details of physical

structures• Uses concept of three-level information

architecture & Data Dictionary– User or application requests manipulation of data

items by name• DBMS uses data dictionary• Maps from attribute name to fields in records in

physical file


• Three – Level Information Architecture– External View

• Describes one user’s view of the database

• Several users or programs can share the same view

• User• Programmers


• Three – Level Information Architecture– Conceptual Schema

• Describes all of the enterprise’s data in the database

• Describes all of the constraints that apply

• Describes logical organization & layout of the database

• Designer


• Three – Level Information Architecture– Internal Schema

• Describes how the data is stored• Describes how the data is accessed• DBA

– Described in the Data Dictionary• Transparent to user


• Three-Level Architecture Supports Program/Data Independence– Change in external view

• Add a new external view for a user• Existing application programs need no

change• Change in a user’s external view

impacts only that user


• Three-Level Architecture Supports Program/Data Independence– Change in conceptual schema

• Add a data item or a new entity• Existing application programs need no

change• An external view may need to be

redefined


• Three-Level Architecture Supports Program/Data Independence– Change in internal schema

• Changing patterns of usage necessitate tuning

• Can create/drop indexes, change storage media

• Existing application programs need no change


• Three-Level Architecture Supports Program/Data Independence– Information hiding

• Because physical storage details are not exposed to user, user is immune to changes in the physical storage


• How does an RDBMS work?– A user issues some request, using a

particular data language– The RDBMS parses the request and

analyzes it– Using the data dictionary, the RDBMS

inspects• The external view of that user• The corresponding external/conceptual

mapping

End 09-09-05


• How does an RDBMS work?• The conceptual schema• The conceptual/internal mapping• The storage structure definition

– The RDBMS generates an optimized query plan & executes it

– Returns the requested data to the user


• DBMSs have Evolved– DBMS began as a computerized record-

keeping system– More specialized, user-oriented packages

have been added– Evolved to a DBMS engine & toolkit

concept• Supports client/server & internet

architectures


• The RDBMS engine & the toolkit functions– RDBMS engine stores, retrieves, &

protects the data• Manages the data structure• Handles transaction processing,

concurrency• Manages security• Supports one standard interface: SQL


• The RDBMS engine & the toolkit functions– RDBMS toolkit manipulates & presents

data• Composed of a set of products &

utilities• Provides a wide variety of user

interfaces• Query languages, report writers, CASE

tools, application development tools


• Client/Server Architecture– Client/Server architecture implements

engine – toolkit in a network• RDBMS engine executes on server

system• Toolkit products & utilities execute on

client workstations


• Client/Server Architecture– Client/Server architecture supports “open

systems”• RDMBS engine interfaces with toolkit

products from multiple vendors• Leads to increased emphasis on

standards: SQL


• Client/Server Architecture– Client/Server architecture made possible by

current technology• Networking• Standards for communications & SQL• Proliferation of software for workstations &

servers– Internet architecture is multi-tiered client/server

• Client/Server interface is through a browser


• Problems an RDBMS Does Not Solve– Coordination among departments of

groups of users– Inadequate training– Missing, incomplete, or out-of-date

requirements– Insufficient time for thorough analysis &

design– Poor application program design


• Problems an RDBMS Does Not Solve– Existing errors in the data

• “Dirty” data, dead data– Duplicate data

• Inconsistent definitions, formats, values– Determining who is responsible for the

data


• Consequences of using an RDBMS– Need for skilled design & management– Training of programmers & users– Changes to existing procedures– Changes to job functions; DBA– Conversion to RDBMS

• Data• Programs


• Consequences of using an RDBMS– Additional or different hardware

configuration• Server in a network

– May be incompatible with other DBMSs– Expensive


• End Section


• See you next time…

IT 20303 The Relational DBMS. Relational Database Theory Building a Database –Begin by developing a conceptual model of the application domain Based on.

Documents

data slide

data access

data dictionary slide

data data manipulation

data security

data structures

tablespaces data

relational database