Discuss the concept of object identifiers (OIDs) in an object DBMS and discuss four different approaches for the representation of OIDs.

What will be an ideal response?

OIDs are used both by application programs for referencing objects and for representing relations between objects. The choice of the type of representation of OIDs can also influence the performance of an ODBMS. OIDs can be represented in different ways (for a review of the different approaches proposed, see Khoshafian and Copeland (1986)).
An OID can be physical or logical. The former contains the actual address of the object, whereas
the latter is an index from which the address of the object is obtained. Different approaches have
been proposed for the representation of both physical and logical OIDs, thereby producing at least
four types of OID.
? Physical address
The OID is the physical address of the object. This representation, normally used by
programming languages, has the advantage of being very efficient, but it is rarely used in
an ODBMS since, if a given object is moved or deleted, all the objects containing its
OID must be modified.
? Structured address
The OID consists of two parts – the first contains the segment number and the page
number, thus making it possible to obtain quickly the address on the disk to be read,
whereas the second part contains a logical slot number, which is used to determine the
position of the object within the page. With this representation, the object can be
relocated within the page simply by changing the slot array, or it can be moved to
another page by inserting its forward address in the slot array.
? Surrogate
The OID is generated by using an algorithm which guarantees its uniqueness (for
example, the time and date, or a monotonically increasing counter). Surrogate OIDs are
then transformed into the object’s physical addresses, normally by using an index.
? Typed surrogates
A variant of the surrogate for representing OIDs involves having both a type identifier
(type ID) and a portion of the object identifier. A different counter generates the object
identifier portion for each type. Thus, the address space is segmented. Moreover, the
type identifier in the object’s OID allows us to determine the object type without
retrieving the object from the disk.