Course Outline

DBMS

|

 |

 Definitions
‡ Data: Meaningful facts, text, graphics,
‡ images, sound, video segments
‡ Database: An organized collection of
logically related data
‡ Information: Data processed to be
useful in decision making
‡ Metadata: Properties /characteristics
that describes data
|

 |

 4hat is a database (DB)?
‡ A collection of data that exists over a
long period of time, often many years¶
‡ Managed through a database
management system

|

 |

4hat is a database management
system (DBMS)?
‡ A powerful tool for creating and
managing [and manipulating] large
amounts of data [(several gigabytes)]
efficiently and allowing it to persist
over long periods of time, safely
‡ Focus on secondary, rather than main,
memory
‡ Powerful, but simple, programming
interface
|

 |

 A Simple Data Management
Problem
Suppose we want to save Names, Phone
Numbers«
‡ Solution 1 (Paper based)
² A blank notebook OR a phone/address book
² Entries recorded by pen, in time order
‡ Advantages
² Cheap, simple, private, reliable, space efficient
‡ Disadvantages
² Hard to search, update, share, expand
² Hard to add information, e.g. email addresses

|

 |

 Another approach
The Traditional File
Processing Environment
Use of Note Pad, Ms. 4ord, MS Excel

|

 |

 DBMS vs. `just a file system'
‡ DBMS's evolved from file systems
‡ file systems also store large amounts of data
over a long period of time in secondary
memory
‡ however, file systems
² can lack efficient access
² have no direct support for queries
² limit organization to directory creation and
hierarchical organization
² have no sophisticated support for concurrency
² do not ensure durability
|

 |

 ACID properties
( All good DBMS's should guarantee these )
‡ Atomicity
² should not be able to execute half of an operation
² either all or none of the effects of a transaction are made permanent
‡ Consistency
² The consistency property ensures that any transaction the database performs will take it from one
consistent state to another.
² there should be no surprises in the world, e.g., gpa > 4.0, balance < 0, cats should never have more
than 1 tail!
² the effect of concurrent transactions is equivalent to some serial execution
² use constraints, triggers, active DB elements (context-free)
‡ Isolation
² Isolation refers to the requirement that other operations cannot access data that has been
modified during a transaction that has not yet completed.
² concurrency control
² transactions should not be able to observe the partial effects of other transactions
² use locks (whole relations or individual tuples?)
‡ Durability
² Durability is the ability of the DBMS to recover the committed transaction updates against any kind
of system failure (hardware or software).
² if power goes out, nothing bad should happen
² once accepted, the effects of a transaction are permanent (until, of course, changed by another
transaction)
² use logs

|

 |

 Applications of database
systems
‡ reservation systems, banking systems
‡ Network simulations / Experiments
‡ record/book keeping (corporate, university, medical), statistics
‡ bioinformatics, e.g., gene databases
‡ criminal justice
² fingerprint matching
² how do you encode `looks like'?
‡ multimedia systems
² require terabytes (1012 bytes) of storage
² tertiary storage devices, e.g., CD, DVDs
² image/audio/video retrieval
² streaming, interactivity
‡ satellite imaging; can require petabytes (1015 bytes) of storage
‡ the web
² client-server and multi-tier architectures
² almost all data-intensive websites are database-driven; IMDB.com is an exception
‡ information integration
² over the web
² legacy systems; must deal with issues of
‡ synonymy: different words having the same meaning, e.g., coffee shop vs. café
‡ polysemy: same word (homonym) having different meanings, e.g., shot
² data warehouses
² data mining (KDD, Knowledge Discovery in Databases), e.g., association rules: `diapers beer'; we pass these on to
the marketing folks
‡ in sum, databases are everywhere!

|

 |

 Three classical data models
‡ hierarchical model
‡ network model
² each tuple is a separate record,
² no separation between logical and physical views
² used record-at-a-time languages
² too low-level
‡ relational model
² most popular and successful model
² de facto standard for databases
² (relational) databases are one of the most popular success stories of
simple theoretical ideas
‡ semistructured data and XML
² semistructured data is self-describing
² web data tends to be semistructured
² in between structured and unstructured data (free text)
² the study of the storage and retrieval of unstructured data is called IR
(Information Retrieval)

|

 |

Main themes of relational database
management systems (RDBMS's)
‡ data stored in a relation (for now, a table), e.g., a simple relation

‡ attributes (columns), tuples (rows, records)

‡ clean separation between logical and physical views, ANSI Sparc

Views

Relations

physical storage

|

 |

 Contd««
‡ gives rise to powerful, yet declarative, relation-at-a-
time query languages, e.g., SQL (Structured Query
Language; pronounced `sequel')
‡ a simple SQL query illustrating the SELECT-FROM-
4HERE construct
² SELECT id FROM Students 4HERE major = 'CPS' AND GPA
> 3.7;
‡ relational query languages (QLs) are declarative
² you specify what you want, not how to get it (à la PROLOG)
² e.g., SQL
‡ closure property

|

 |

 How can we study database
systems?
‡ design of databases, i.e., how do you
structure your data in a database?
² entity-relationship (E/R) model
² relational model

‡ database programming
² how do you use a DBMS?
² study (query languages) such as SQL

‡ database system implementation, i.e., how

|

 |