A Database Management System (DBMS) is a combination of hardware and software that enables one or more computers or servers to store, organize, and retrieve data efficiently.
In practice, a DBMS handles data storage as well as data processing, security, and integrity. Information is structured in tables, where each row represents an element or object and each column represents an attribute or characteristic.
1. Database Management System vs. ERP System
A Database Management System (DBMS) is software that manages the storage and retrieval of data. Its role is to ensure that data remains reliable and accessible at all times.
An Enterprise Resource Planning (ERP) system is software that manages all of an organization’s business processes, including product planning, procurement, manufacturing, sales, and customer service.
The ERP system provides management with insights on company performance, enabling better decision-making. It acts as a source of decisional computing, hosted in a Decision Support System (DSS), also called a Business Intelligence System.
ERPs also rely on Electronic Data Interchange (EDI). In e-commerce transactions, the customer sends a message to the supplier containing product specifications and pricing information — automatically and transparently.
2. How do Database Management Systems work?
A database is a collection of information stored in a computer system, structured as tables with rows and columns. The Database Management System organizes that data both physically and logically, similar to how an Excel spreadsheet presents information.
The technical architecture of a database refers to the physical layout of data storage, including the location of the data on the hard drive or other storage device.
In contrast, cloud architecture refers to how data is accessed over a network. Information is distributed across multiple servers and reachable via the Internet or a private network (VPN). This domain is known as Cloud Computing.
3. Local DBMS vs. Cloud DBMS
A local DBMS is installed on servers within the company’s network and accessible only to its employees.
A cloud DBMS is hosted in a cloud provider’s data centers and reachable over the Internet from any connected device.
Cloud databases are typically more expensive, since the provider handles infrastructure, maintenance, and support. In return, they offer scalability (handling growing data and users without new hardware), flexibility (customizable to the organization’s needs), availability (accessible anywhere, anytime), and security (encryption, authentication, and backups).
Local databases are cheaper because there is no third-party infrastructure to pay for. However, they offer limited scalability, flexibility, and availability, and remain confined to the local network.
In short, the choice depends on the organization’s needs and budget: local databases suit small businesses with modest workloads, while cloud databases fit larger organizations with heavier demands.
Follow the next article to learn more about ETL and data integration.
4. What are the different types of databases?
There are many types of database management systems, each with their own advantages and trade-offs. Some store data in tables (relational), others in documents, graphs, cubes, or even raw vectors. The right choice depends on the workload — transactional, analytical, real-time, search, or AI.
Relational databases — also called Online Transactional Processing or OLTP databases — execute analytical queries more slowly than OLAP systems, but remain more flexible for schema changes. SQL is used to query relational databases, whereas OLAP cubes are queried with Multidimensional Expressions (MDX).
Below is an exhaustive overview of the main database categories and the leading vendors in each.
4.1. Relational databases (RDBMS / SQL)
Data is organized into tables linked by keys, with strong ACID guarantees and SQL as the standard query language.
- Commercial: Oracle Database, Microsoft SQL Server, IBM Db2, Teradata
- Open source: PostgreSQL, MySQL, MariaDB, SQLite
- Cloud-native: Amazon Aurora, Google Cloud SQL, Azure SQL Database
4.2. Hierarchical databases
Data is organized in a tree-like structure with parent-child relationships. This model is well-suited for clear hierarchies such as organizational charts or file systems.
- IBM IMS, Windows Registry
4.3. Network databases
An evolution of the hierarchical model, where records can have multiple parent-child links, forming a graph-like structure.
- IDMS, Raima Database Manager
4.4. Object-oriented databases
Information is stored as objects, which combine both data and methods. Such systems fit applications that require complex data relationships and fast access to data.
4.5. NoSQL databases (overview)
Designed for large-scale, distributed data storage, NoSQL systems do not rely on the traditional SQL language. They handle unstructured or semi-structured data such as social media posts, log files, and sensor data.
NoSQL splits into four main subtypes, detailed in sections 4.6 to 4.9.
4.6. Document-oriented databases
Records are stored as documents (typically JSON or BSON), each containing nested keys and values. Such systems are ideal for content management, catalogs, and user profiles.
4.7. Key-value databases
Information is stored as simple pairs of keys and values, offering very fast lookups. Typical use cases include caching, session management, and real-time analytics.
4.8. Wide-column (column-family) databases
Records are organized into rows and dynamic columns grouped into column families. This model scales horizontally and handles huge volumes of data across distributed clusters.
4.9. Graph databases
Information is represented as nodes (entities) and edges (relationships). This structure fits applications such as fraud detection, recommendation engines, and social networks.
4.10. NewSQL databases
A modern category combining the scalability of NoSQL with the ACID guarantees of relational databases.
4.11. Columnar databases
Unlike row-based systems, columnar databases store values column by column on disk, which dramatically accelerates analytical queries and aggregations. This approach powers most modern data warehouses.
4.12. OLAP databases (multidimensional cubes)
Classic Online Analytical Processing systems based on pre-aggregated multidimensional cubes, queried with MDX.
- SSAS Multidimensional, Oracle Essbase, Oracle OLAP Option, IBM Cognos TM1 / Planning Analytics, Apache Kylin, Mondrian (Pentaho)
4.13. Tabular databases (xVelocity / VertiPaq)
A Microsoft implementation of columnar in-memory analytics, queried with DAX. The engine is shared across SSAS Tabular, Power BI, and Azure Analysis Services.
- SSAS Tabular, Power BI datasets, Azure Analysis Services
4.14. Cloud data warehouses (modern OLAP)
Distributed columnar warehouses optimized for analytical SQL queries at petabyte scale.
4.15. Real-time / streaming OLAP databases
Built for low-latency analytics over high-velocity event streams.
4.16. Time-series databases
Optimized for time-stamped data points such as IoT sensor readings, server metrics, or financial market data. High-volume writes and efficient time-range queries are core features.
4.17. Spatial / geographic databases
Built to store and query geographic and geometric data — maps, locations, routes — powering GIS, logistics, and location-based services.
4.18. Vector databases
A newer category designed to store and search high-dimensional vector embeddings produced by machine learning models. Vector databases power semantic search, recommendation systems, and AI / RAG applications.
4.19. Search-oriented databases
Specialized engines built for full-text search, faceted navigation, and log analytics.
4.20. In-memory databases (general purpose)
Data is held in RAM rather than on disk, which removes I/O overhead and enables real-time processing for financial systems, online gaming, and live analytics.
- Commercial: SAP HANA, Oracle TimesTen, Oracle Database In-Memory, SingleStore, Kdb+, SQL Server In-Memory OLTP
- Open source: Redis, Memcached, Apache Ignite, Hazelcast, VoltDB, Aerospike, DuckDB
- Cloud-managed: Amazon ElastiCache, Amazon MemoryDB for Redis, Google Memorystore
4.21. Embedded databases
Lightweight engines that run inside an application process rather than as a separate server. Common in mobile apps, IoT devices, and edge computing.
4.22. Multi-model databases
A single engine supporting several data models — document, key-value, graph, relational — within the same system.
4.23. XML databases
Specialized in storing, querying, and indexing XML documents using XQuery and XPath.
4.24. RDF / triplestore databases
Built for semantic web data stored as RDF triples (subject-predicate-object) and queried with SPARQL.
4.25. Ledger / blockchain databases
Immutable, cryptographically verifiable databases used for audit trails and decentralized applications.
Learn more about XML for Analysis (XMLA)
Conclusion on Database Management Systems
A Database Management System (DBMS) allows organizations to store, manage, and retrieve data efficiently, while keeping it reliable, secure, and accessible at all times.
The database landscape has expanded far beyond traditional relational systems. Today’s stack ranges from OLTP and OLAP engines to NoSQL, NewSQL, time-series, vector, and ledger databases — each tailored to specific workloads.
When selecting a DBMS, key factors include the industry and business needs of the organization, the volume and shape of the data, the financial budget, and the skills and experience of the IT team.
Be the first to comment