Issue link: https://resources.awscloud.com/i/1496542
©2020 IDC #US47078920 4 TABLE 1 Database Models and Typical Workloads Model How the Model Is Used Typical Workload Relational Schematic, table-oriented data management in support of ACID (atomicity, consistency, isolation, durability) transactions and random, orthogonal queries Back-office accounting and record-keeping capture of calculation data, ACID transactions, random orthogonal query either ad hoc or for reports; for large and complex analytic databases, RDBMS supports data warehouses Key-value Non-schematic data grouped in blocks functionally for rapid, random insert and retrieval; data format is under program control High-touch customer experience (CX) applications, ecommerce applications, and user session management for mobile apps and gaming; also, for any application requiring high scalability and the flexibility of NoSQL and does not require complex query support Document Holds data in a standard document format such as JSON or XML, supporting more complex data structures Operational applications with dynamic data requirements, including data definitions, and supporting a wide variety of applications that require management of blocks of formatted data Graph Collecting and analyzing data capturing relationships among many entities, such as persons and locations Can be used for text semantics capture and analysis for search or cataloguing purposes or for navigation or pattern analysis, such as in mobile tracking systems or for fraud detection Wide column store Holds data in defined column groups, enabling multiple applications and users to share the data High throughput operational applications that require speed and scalability along with some random query capability and not as much definitional flexibility as is delivered by a document database In-memory data store Live data sharing among applications or running processes in the same application on different servers; expands available data IoT data capture and analysis, log data analysis, capture of streaming data, recording of changed data, and managing of events Specialized data (e.g., time series and ledger database) Capturing and analytic functions for time series, spatial, or ledger data Analysis platform for log data or location data Source: IDC, 2020 In some cases, where one of these models dominate but the others are needed in the same database, a multimodel DBMS is called for. In most other cases, a purpose-built DBMS aimed at a specific model is preferred. A collection of applications may require some combination of such purpose-built databases, with each assigned to the type of application to which it is best suited.