Switch to: V10V9V8V7V6V5

The Network Database Model

The network database model first evolved in parallel to the hierarchical database model. One early, very popular network database was the Integrated Database Management System (IDMS).

Here is a quick introduction of the network database model - what makes it different or similar to other types of databases:

  • Fields are the smallest units of the database structure. Each field also an unique name.
  • Records are units of database processing ( records are memorized, deleted, read); records are aggregates which are not in a structure of other aggregates.
  • Links are hierarchical relations between records of two types. A record of the first type is the owner of the relation, and a record of the second type is and owned member of another.

For example, consider records for a PERSON and a PHONENUMBER in a phone directory. PHONENUMBER is owned by the the record of the PERSON.

Data aggregation is a nominate collection of fields or of other data aggregates. For example, ADDRESS would include the following fields: INDEX, HOUSE NUMBER.

The Network Data Model provides a relation between records using direct pointers. Using direct pointers is the fastest method of traversing a network database - what was known at one time as “walking the set”.

Linked records are organized into lists. Every owner-record has a pointer to the first owned record in a list. Every owned record has a NEXT pointer: it points to another owned record from the list ( see pict.1).

Different implementations can represent lists by different ways: from a single-linked list to a double-linked list with a pointer to owner in every owned record (see pict.2). Since pointers in lists are stored directly in the record it makes changing of the structure much harder but also extremely fast.

A Network DBMS automatically creates pointers to organize lists of linked records. It is easy to see that double-linked lists would use:

3 * 4 = 12 byte for every owned record and 
2 * 4 = 8 bytes for parent records. 

If N relations exist for a record then we get an overhead N * 12 bytes.

In practice, many network DBMS did not allow changes to structures. An interesting way to help understand navigation in a network DBMS is the World Wide Web. The World Wide Web consists of links between web pages to other related web pages - circular references are a common occurance.