Classification of Database Management Systems

Database Management Systems (DBMSs) can be categorized using several criteria, the most important of which is the data model they use. The relational data model forms the basis of most modern commercial systems and is implemented through SQL-based DBMSs. Other models—such as the object, object-relational, and various NOSQL data models—also exist and are used in specialized environments. NOSQL systems, popular in big data applications, follow different approaches including key-value, document, graph, and column-based data models. Older systems that rely on hierarchical or network models still support many legacy applications.

Relational DBMSs have gradually incorporated features from object databases, leading to object-relational systems that blend relational storage with object concepts. Some experimental systems use the XML data model, which represents data in hierarchical tree structures. While native XML databases exist, many commercial relational DBMSs simply add XML interfaces or storage features.

A second classification criterion is the number of users.

• Single-user systems support only one user at a time, often on personal computers.

• Multiuser systems support concurrent access by multiple users and represent the vast majority of DBMS installations.

A third criterion is the distribution of data.

• In a centralized DBMS, the database and DBMS software reside at a single site.

• In a distributed DBMS (DDBMS), the data and software are spread across multiple connected locations.
  Big data environments often involve large-scale distribution with replication across numerous sites to ensure reliability. Distributed systems may be homogeneous (using identical DBMS software at all locations) or heterogeneous (using different DBMSs, requiring middleware). A loosely coupled integration of multiple autonomous databases forms a federated or multidatabase system.

A fourth criterion is cost, though this can vary widely. Open-source systems such as MySQL and PostgreSQL are available at no cost, sometimes with optional paid support. Commercial DBMSs may be distributed as personal editions, evaluation copies, or large enterprise systems sold through licensing schemes—ranging from per-user licenses to full site licenses. Advanced features, such as parallel processing, replication, mobile capabilities, and data warehousing, often add to the overall cost, and enterprise-level installations can be very expensive.

DBMSs can also be classified by the types of access paths they support. Some systems rely on specialized file structures, such as inverted files, to speed up retrieval.

Finally, DBMSs may be general-purpose or special-purpose. Special-purpose systems are designed for specific, high-performance applications—such as airline reservation systems or telephone directories—and fall under online transaction processing (OLTP) systems. These systems are optimized to process large volumes of concurrent transactions with minimal delay.

Relational, object, object-relational, NOSQL, XML, hierarchical, and network models represent the major data model families. Each offers distinct ways of organizing and accessing data, and their evolution continues to shape modern database technologies.