Data Warehouse Dimension Modeling

Dimension Modeling Method

Dimension modeling organizes information into structs, which typically correspond to the query methods that analysts want to use for data warehouse data. How much food sales were in the northwest in the third quarter of 1999. Represents the use of three dimensions (product, geography, time) to specify the information to be summarized. Star mode is widely used, in order to do a lot of preprocessing for each dimension, such as according to the dimensions of the advance statistics, classification, sorting and so on . Through these preprocessing, can greatly enhance the processing capacity of the Data warehouse. Especially for the 3NF modeling method, star mode has a significant advantage in performance. At the same time, the other advantage of dimension modeling is that the dimension modeling is very intuitive and closely surrounds the business model, which can directly reflect the business problem in the business model. you do not need to undergo a special abstraction to complete dimension modeling. This is also the advantage of dimensional modeling. The advantages of dimension modeling can be summarized as follows: A the dimension modeling is a predictable standard framework . Allows database systems and end-user query tools to generate powerful assumptions about the data, which work primarily on performance and performance. b The predictable framework of the star join pattern can tolerate unpredictable user behavior changes. C has a very good scalability to accommodate unpredictable new data sources and new design decisions. It is convenient to add new analytical dimensions and facts without changing the size of the model, without overloading the data and not needing to be coded to accommodate the new changes. Better scalability means that all previous applications can continue to run without producing a different result. However, the shortcomings of dimensional modeling are also very obvious, because of the need for a large number of data preprocessing before building a star pattern, resulting in a large number of data processing work. Moreover, when the business changes and needs to be redefined for the dimension, the preprocessing of the dimension data is often required. In these and processing processes, often lead to a large number of data redundancy. The disadvantage of another dimension modeling method is that if we only rely on the simple dimension modeling, we can not guarantee the consistency and accuracy of the data source, and it is not particularly suitable for the dimension modeling method at the bottom of the Data warehouse.