Architecture of Btrieve

Btrieve is a database developed by Pervasive Software. The architecture of Btrieve has been designed with record management in mind. This means that Btrieve only deals with the underlying record creation, data retrieval, record updating and data deletion primitives. Together with the MicroKernel Database Engine it uses ISAM, Indexed Sequential Access Method, as its underlying storage mechanism.

Btrieve is essentially a database that uses keys and indexes to organise data. However, the file structure itself is largely built around smaller units of data, called "pages" in Btrieve. Though the structure has changed over the various versions of Btrieve, the file structure still revolves around a File Control Record (FCR) — which defines the configuration of pages — and pages in the Btrieve file that contain data. Historically, Btrieve used "physical pages", or pages that were located at fixed positions in the file. Beginning with version 6.0 "logical pages" started to be used, which were mapped to page allocation tables (PATs) — this allowed Btrieve to change their record update technique from what was later known as "pre-image paging" to a technique called "shadow-paging".

Btrieve is committed to backward compatibility, as versions of Btrieve until version 6.15 use a standard file format and, until Btrieve 6.0 was released, were completely backwards compatible. Btrieve 6.0 introduced new features and had to break compatibility with older versions of the software to implement more advanced features. The API likewise remained backwards compatible, with only one feature (split files to separate media) being dropped. At one point, Btrieve's former CEO Ron Harris stated that "The version 1.0 API is still supported in version 6.15, and we're going to keep it forever!" (Kyle, pg 11).

Pervasive initially used the term "navigational database" to describe Btrieve, but later changed this to "transactional database". The use of the term navigational database was unusual because a navigational database uses "pointers" and "paths" to navigate among data records, and these pointers are contained in the record itself; ISAM, which is the fundamental structure of Btrieve, uses a secondary index table to store these pointers to decrease search times. Thus, the two types of database are different, and may or may not explain why Pervasive started using different terminology for classifying their database. (Note: This is not strictly correct. A navigational database is one in which the logical access to the data in the database is done via the application-level interface or API, It is navigational in the sense that logical relationships are traversed by the application code "navigating" its way through the database. What physical techniques are used to accomplish this, i.e., ISAM, embedded pointers, etc., is almost irrelevant to the discussion. By contrast, a relational database does not provide the application layer with any way to "navigate" through the logical database structure and instead provides a set-level interface for selecting, aggregating and joining data. Relational databases may also use a variety of physical techniques to access data including the same mentioned above, but the important aspect of being "relational" is that the data is accessed relationally, i.e., vie a set query model rather than a navigational model.)

...
Wikipedia