Objects were developed by Xerox Corporation's
Palo Alto Research Center in the early 1970's. In Object-Oriented databases,
the focus is on objects, not functions. Object-oriented (OO) technology
requires a totally new mindset; developers create models that are mapped to real
world processes.
The object-oriented approach emphasizes a more
natural representation of the data. In today's environment, the data models are
more demanding. They need to handle audio, video, text, graphics, etc. These
requirements demand a more flexible storage format than hierarchical, network,
and relational databases can provide. Only object-oriented databases will be
able to support this kind of demand.
Three of the fundamentals of OO technology are
classes, objects, and messages. A class describes a group of objects that have
common relationships, behaviors, and also have similar properties. An object
encompasses related data and functions into a completely self contained
package. Because objects are "stamped" out from a common class definition, each
instance of an object within a class inherits all of the same behaviors and data
definitions. The heart of an object-oriented database is object persistence,
and it is the process of storing and retrieving objects that comprises the heart
of object data management.
A database object may have a different OID each
time that it is loaded onto the heap (memory), and the OODBMS must manage and
assign the OID's according to the memory location after the object has been
retrieved from disk..
Lets take a look at how behaviors or methods are
invoked within an object database. For example, assume that we have an object
of class order with an object id of 123. The object oriented call to compute
the total of order 123 might look like this:
order.compute_order_total(123);
In this case, the message compute_order_total is
being sent to the order class definition along with the oid that will be the
target of the message. A method by the name compute_order_total will then be
executed against order 123 and the desired data manipulation will take place.
In distributed systems, messaging becomes more
important because the class definitions that contain the behaviors may reside on
geographically remote processors. For more information on distributed message
processing see chapter 7, Distributed
object technology and database management.
Each time an object is created, a unique OID
(object identifier) is added to the OODBMS (object-oriented database management
system) identifier table. When an application references an object via its OID,
the OODBMS converts the OID into a virtual memory address. This means that the
object can be found quickly regardless of where the object is stored. For
example, local memory, a remote hard disk, or a device in a networked system.
The OID is independent of the value of an object or any data contained in the
object. You can change any or all of the variables in an object and the system
will still access the correct object. An OID once assigned to an object by the
system, lasts the lifetime of an object. An OID is much different from the
concept of a key in a relational database. In a relational database, the key,
is defined by one or more values of the fields of a table. The key in a
relational table can be modified. In an OODBMS, the identity of an object is
system generated and cannot be changed. Any two objects are different because
of their OID, even if they have the same value.
Let's show how OODBMS relate to RDBMS. Object
classes correspond to relational tables, object attributes correspond to
columns, and instances of objects correspond to rows in a table. Objects are
related to each other by the OID, this corresponds to primary and foreign keys
in the relational model. For more information on this technique, see Chapter 9,
Interfacing an object-oriented
application with a relational database.
The object technology model also provides for
the encapsulation of data and operations inside an object. As long as the
external interfaces remain unchanged, developers can modify encapsulated data
and operations in an object without affecting other objects. Encapsulation
makes the data and procedures private within an object.
As you can see, OODBMS have many new terms,
concepts and ways of doing things. The potential benefits of the object
databases are faster development, higher quality, easier maintenance, and
increased scalability and adaptability. In addition the object databases claim
to allow the MIS personnel to better model the real-world, resulting in better
information structures and reduced costs for systems. Potential concerns with
object databases are
1.
lack of maturity
2.
the need for better tools
3.
the need for standards
4.
the availability of
qualified personnel
5.
the costs of data
conversion
The object-oriented approach is not only
concerned with data storage, data relationships, and easy data access, but also
the behavior storage. This fundamental difference will revolutionize the way
that databases function. Rather than having all of the DML in hundreds of
external application programs, the DML will be tightly coupled with the class
definition for their associated objects, and will reside in the DBMS along with
the data. This implies that business processes and data will share a common
ground.
Object Persistence
Object persistence is one of the most
fundamental services provided by an OODBMS. First let's define what
persistence means. A persistent object is an object that is always available
and an object's state will remain unchanged from one invocation to the next, as
opposed to an object that is no longer available.
Objects that are no longer available are called
transient objects. Depending on the application, some objects may need to
change their state from transient to persistent, and the OODBMS will manage the
conversion from persistent to transient. The OODBMS assigns each object an
object identifier (OID), this unique identifier is used primarily to establish
relationships between persistent objects. OODBMS provide the recovery services
that ensure persistence objects survive any kind of system failure. The
ORB(Object Request Broker) will send a request for a particular object to the
OODBMS, which will use the OID to retrieve and invoke the object. There are two
methods that the OODBMS uses to access persistent objects, virtual memory
address pointers and hash tables. A persistent object will always be ready to
be invoked, and its state can and will be preserved and survive any kind of
system failure.
Remember, persistent objects are stored on disk
and transient objects exist in RAM memory. In an OODBMS, an object may traverse
between states.
The following are some of the actions that a
OODBMS will perform for object persistence, the most important, of course, being
the management of objects as they are retrieved from disk and assigned new OID's
in RAM memory:
1.
will assign space in persistent
storage
2.
will automatically increase space
when needed
3.
will manage the free space in
persistence storage
4.
will interface the I/O system with
persistent storage
5.
will ensure that persistent objects
can be recovered
6.
will manage buffers
7.
will map between the physical
addresses and the OIDs.
and include the URL for the page.
