The Multi-dimensional database model (sometimes called the "Post-relational" "Multi-valued" or "Pick" model) offers significant advantages over the traditional "flat" or "relational" databases:

Multivalued data

This is the area in which the two models differ the most. multi-dimensional databases allow multiple (or even no) occurrences of any data field. This allows them to mirror the real world more closely.

Consider a personnel file. The records, or "rows", in this file may include a telephone number field. An employee may have one telephone number, many numbers or no telephone number at all. And the telephone numbers may vary in length.

If you were using a relational database you'd have to provide space in each row for an arbitrary number of occurrences of the telephone number field, and each occurrence would be of a fixed, arbitrary length.

And if, at a later stage, you were to increase the number of occurrences, or to change the length of one of these fields you'd probably have to modify every program that accessed that file.

But if you were using a multi-dimensional database (such as UniVerse) the system will automatically allow as many occurrences of that field as each record (row) requires, and allow each occurrence to be as long as it needs to be.

Similarly records (rows) are considered dynamic both in terms of length and in terms of the number of fields.

And the programming languages supplied with the DBMS include functions to handle variable length records and multiple values - thus reducing program maintenance if a record layout is amended.

More efficient use of disk resources

Consider the above example: In the traditional scheme of things you'd have to provide for n occurrences of that field, and fix the length of the field.

And that space (n times the field length) is used for each row whether the fields contain any data or not.

But in the multi-dimensional scenario all that is written to disk is the actual data (with no padding) and a single field delimiter - no unnecessary space is occupied. This only saves a few bytes per record, that's true. But this saving becomes significant in a large database.


Each record (or "row") in a file (or "table") is identified by a unique key field. And all records are directly accessible by this key.

There's no need to parse the entire table. If you have the key you can efficiently and directly access the record that you want.

And access via this primary key does not require any indexing - although this facility is available to provide alternate record keys.

Date Handling

This family of DBMS uses a special algorithm for encoding dates for internal storage. This results in dates that, when stored on disk, occupy less space than dates stored in the MMDDYY (or similar) format; yet can be converted into various formats (23-01-97; 23 January 1997, 01/12/1997 etc.) for display. This date handling scheme also greatly reduces the problems associated with working with dates that fall within different centuries.

Easy ad hoc database querying

Multi-dimensional DBMS offer an easy to use, English-like language for querying and reporting from the database.

An especially useful feature is the "select list". In "relational" terms a select list is a list of rows from a table that meet user specified criteria. Once the list is established subsequent processing or queries are carried out on the selected rows only.

Support of open standards

Multi-dimensional DBMS now integrate with most of the de facto "open" standards and programming tools such as ANSI SQL, ODBC, and Visual Basic. They thus interact easily with modern Windows based products.

Portability and scalability

Multi-dimensional DBMS are available for Windows NT and all major commercial UNIX variants. And applications and data are easily ported across hardware and software platforms.

Your multi-dimensional system will move and grow with your business.

| Links | Home |