multidimensional database

Multi-dimensional database (Multi Dimensional Database, MDD) can be simply understood as: storing data in an n-dimensional array, rather than storing it in the form of records like a relational database. So it has a lot of sparse matrices and people can look at the data through a multidimensional view. Multidimensional database adds a time dimension. Compared with relational database, its advantage is that it can improve data processing speed, speed up response time, and improve query efficiency.
　　There are currently two MDD OLAP products: MOLAP based on multidimensional databases and ROLAP based on relational databases. ROLAP has established a new system, the star structure .
　　MDD does not have a recognized multi-dimensional model, nor does it have a standard way of getting data (eg SQL, API, etc.) like the relational model. MDD-based OLAP products also vary greatly depending on the content of decision support.
　　At the low end, users use single-user or small LAN-based tools to view multidimensional data. The functionality and usefulness of these tools may be quite good, but they do not have all the features of OLAP due to their limited size. These tools use a hypercubic structure that constrains the model to n-dimensional morphology. When the model is large enough and the sparse data is not well controlled, such a model will be vulnerable. These tools use databases that are measured in megabytes rather than gigabytes, so they are read-only and have limited computational complexity.
　　At the high end, OLAP tools with 4GL provide a complete development environment, statistical analysis, time series analysis, financial reporting, user interface, multi-tier architecture, charts, and many other features. Although different OLAP tools use their own multidimensional databases, they also utilize relational databases as storage media to varying degrees. Because relational databases and OLAP tools are processed simultaneously on high-end servers, speed and efficiency are still high.
　　Pure multidimensional database engines were also developed. Although these tools lack 4GL and an adequate development environment, they have more complex databases than those used by high-end MDD tools. These tools also have functions such as statistical analysis, financial analysis, and time series analysis, and have their own APIs that allow them to be open to front-end development environments.
　　MDD can provide excellent query performance. The information stored in the MDD has a more detailed index than the information in the relational database and can be resident in memory. MDD information is stored in the form of an array, so it can update the data without affecting the index. Therefore MDD is very suitable for read and write applications.