Database Management Course Content Systems • Introduction • Database Design Theory • Query Processing and Optimisation Winter 2003 • Concurrency Control • Data Base Recovery and Security CMPUT 391: Object Oriented Databases • Object-Oriented Databases • Inverted Index for IR Dr. Osmar R. Zaïane • XML • Data Warehousing • Data Mining • Parallel and Distributed Databases University of Alberta Chapter 23 of • Other Advanced Database Topics Textbook Dr. Osmar R. Zaïane, 2001-2003 1 Dr. Osmar R. Zaïane, 2001-2003 2 2 Database Management Systems University of Alberta Database Management Systems University of Alberta Object-Oriented Databases Objectives of Lecture 6 Object-Oriented Databases • Shortcomings of Relational Databases • Discuss limitations of the relational data • The Concept of Object data Model model. • Introduce object databases, databases that • Object-Oriented Database Systems handle complex data types. • Object-Relational Database Systems • Understand the difference between object- • CORBA oriented databases and object-relational databases. • (By no means comprehensive) Dr. Osmar R. Zaïane, 2001-2003 Dr. Osmar R. Zaïane, 2001-2003 3 4 Database Management Systems University of Alberta Database Management Systems University of Alberta
The Need for a DBMS Manipulating New Kinds of Data • On one hand we have a tremendous increase • A television channel needs to store video in the amount of data applications have to sequences, radio interviews, multimedia handle, on the other hand we want a reduced documents, geographical information, etc., and application development time. retrieve them efficiently. – Object-Oriented programming • A movie producing company needs to store – DBMS features: query capability with movies, frame sequences, data about actors and optimization, concurrency control, recovery, theaters, etc. (textbook example) indexing, etc. • A biological lab needs to store complex data • Can we merge these two to get an object about molecules, chromosomes, etc, and retrieve parts of data as well as complete data. database management system since data is getting more complex? • Think about NHL data and commercial needs. Dr. Osmar R. Zaïane, 2001-2003 5 Dr. Osmar R. Zaïane, 2001-2003 6 Database Management Systems University of Alberta Database Management Systems University of Alberta Shortcomings with RDBMS What are the Needs? • Images • Supports only a small fixed collection of relatively • Video simple data types ( integers, floating point numbers, date, strings ) • No set-valued attributes (sets, lists,…) • Multimedia in general • No inheritance in the Is-a relationship • Spatial data (GIS) • No complex objects, apart from BLOB (binary • Biological data large object) and CLOB (character large object) • CAD data • Impedance mismatch between data access language (declarative SQL) and host language • Virtual Worlds (procedural C or Java): programmer must • Games explicitly tell how things to be done. • List of lists ÿ Is there a different solution? • User defined data types Dr. Osmar R. Zaïane, 2001-2003 Dr. Osmar R. Zaïane, 2001-2003 7 8 Database Management Systems University of Alberta Database Management Systems University of Alberta
Existing Object Databases DBMS Classification Matrix • Object database is a persistent storage manager for objects: Relational Object-Relational Query – Persistent storage for object-oriented programming DBMS DBMS languages (C ++ , SmallTalk,etc.) – Object-Database Systems: Object-Oriented DBMS • Object-Oriented Database Systems: alternative to relational No Query systems File System • Object-Relational Database Systems: Extension to relational systems • Market: RDBMS ( $8 billion), OODMS ($30 million) world-wide Simple Data Complex Data • OODB Commercial Products: ObjectStore, GemStone, Orion, etc. Dr. Osmar R. Zaïane, 2001-2003 9 Dr. Osmar R. Zaïane, 2001-2003 10 Database Management Systems University of Alberta Database Management Systems University of Alberta Object Data Model Object-Oriented Databases • The object data model is the basis of object- oriented databases, like the relational data model • Shortcomings of Relational Databases is the basis for the relational databases. • The Concept of Object data Model • The database contains a collection of Objects • Object-Oriented Database Systems (similar to the concept of entities) • Object-Relational Database Systems • An object has a unique ID (OID) and a collection • CORBA of objects with similar properties is called a class. • Properties of an object are specified using ODL and objects are manipulated using OML. Dr. Osmar R. Zaïane, 2001-2003 Dr. Osmar R. Zaïane, 2001-2003 11 12 Database Management Systems University of Alberta Database Management Systems University of Alberta
Abstract Data Type Properties of an Object • One key feature of object database systems is the possibility for the user to define • Attributes : atomic or structured type (set, arbitrary new data types. bag, list, array) • A new data type should come with its • Relationships : reference to an object or set associated methods to manipulate it. The of such objects. new data type and its associated methods is called abstract data type (ADT). • Methods : functions that can be applied to objects of a class. • DBMS has built-in types. • How does the DBMS deal with new data types that were never seen before. Dr. Osmar R. Zaïane, 2001-2003 13 Dr. Osmar R. Zaïane, 2001-2003 14 Database Management Systems University of Alberta Database Management Systems University of Alberta Inheritance Encapsulation •A value has a type •An object belongs to a class • Type (class) hierarchy • Encapsulation = data structure + operations – System permits the definition of new types based on other existing types • It is the main characteristic of object-oriented – A subtype inherits all properties of its supertype languages. • Class hierarchy • The encapsulation hides the abstract data type – A sub-class C’ of a class C is a collection of objects such that each object in C’ is also an object in C. internals. ADT= opaque type. – An object in C’ inherits all properties of C • The DBMS does not need to know how the • may change the behaviour of some methods (overloading/overriding of methods) ADT’s data is stored nor how the ADT’s • typically adds additional attributes and methods • Multiple inheritance (inherits from more than just one methods work. DBMS only needs to know the superclass) available methods and how to call them • Selective inheritance (inherits only some of the properties of a superclass) (input/output types of the methods) Dr. Osmar R. Zaïane, 2001-2003 Dr. Osmar R. Zaïane, 2001-2003 15 16 Database Management Systems University of Alberta Database Management Systems University of Alberta
Common Structured Types Objects, OIDs, and Reference Types • Type constructors are used to combine atomic types and user defined types to create more • An object has an identity and the system complex structures: can generate an object identifier (OID) for • ROW(n 1 , t 1 , …n n ,t n ) : tuple of n fields objects which is unique in the database • listof(base): list of base-type items across time • ARRAY(base): array of base-type items • Reference types - REF(basetype) - have • setof(base): set of base-type items without object ids as values, i.e., an object of type duplicates REF( basetype ) is basically a “pointer” to an • bagof(base): multiset of base-type items object of type basetype . Not all collection types supported by all systems Dr. Osmar R. Zaïane, 2001-2003 17 Dr. Osmar R. Zaïane, 2001-2003 18 Database Management Systems University of Alberta Database Management Systems University of Alberta Object-Oriented Databases Object-Oriented Databases • OODBMS aims to achieve seamless integration • Shortcomings of Relational Databases with an object-oriented programming language such as C++, Java or Smalltalk. • The Concept of Object data Model • OODBMS is aimed at applications when an • Object-Oriented Database Systems object-centric view point is appropriate. • Object-Relational Database Systems (occasional fetch from object repository) • CORBA • No efficient implementations for DML. There are no good optimizations for a query language such as OQL in OODBMSs today. Dr. Osmar R. Zaïane, 2001-2003 Dr. Osmar R. Zaïane, 2001-2003 19 20 Database Management Systems University of Alberta Database Management Systems University of Alberta
Recommend
More recommend
