SSAS Series--"02" Multidimensional Data (Dimension objects)

Original: SSAS Series--"02" Multidimensional Data (Dimension object)

1. What is dimension?

In mathematics, it is called a parameter, which is the number of independent space-time coordinates in physics. 0-Dimensional is a point, 1-dimensional is a line, 2-dimensional is a long and wide (or curved) area, 3-dimensional is 2-dimensional plus the height to form the volume surface. In physics, Time is the fourth dimension, unlike three spatial dimensions, it has only one and can only move forward.

Figure one on the concept of dimensions in MSDN

2, the contents of the server object?

Includes data sources, data Views, database objects, security roles, and assemblies. The first 4 are not difficult to understand. With regard to the concept of assemblies, it is also necessary to reinforce that assemblies can be of two different types: COM and CLR. CLR assemblies are assemblies that are developed using. NET Framework programming languages such as C #, Visual Basic. NET, and managed C + +. To be blunt, you can use the C # language to programmatically manipulate server objects.

3. What is the content of the dimension object?

Simple Dimension objects consist of basic information , attributes , and hierarchies . The basic information includes the name of the dimension, the type of the dimension, the data source and storage mode, and so on. property to define the actual data in the dimension. Properties do not have to belong to hierarchies, but hierarchies are generated by attributes. Hierarchies can not only create ordered lists of levels, but also define how users can browse dimensions. Simply put, just like the key-value pairs in the Session object, here is the key (basic information), Value (property), but there may be a certain relationship here before the key, so a more hierarchical structure.

the structure of the dimension

A, the simplest--star-shaped structure

Features: Each dimension is based on a dimension table that is linked directly to the fact table through a primary key-foreign key relationship. Not much to say, direct.

Figure Two-star structure

b, the most complex--snowflake structure

Figure three snowflake type structure

Storage of dimensions

A, ROLAP

The data that uses the ROLAP dimension is actually stored in the table that defines the dimension. low relative query performance.

B, MOLAP

Data for dimensions that use MOLAP is stored in a multidimensional structure in an analysis Services instance. High relative query performance.

C, HOLAP

A combination of ROLAP and MOLAP storage methods.

4. How is the dimension landed?

The dimension itself is a very abstract thing, and it is linked to the database table through attributes. When it comes to attributes, it becomes more and more enlightened that the concept of attributes is ubiquitous in OO concepts. The dimension is implemented as a collection of attributes that are provided by the columns of the data table in the data source, one or more columns, and multiple columns exist. So find a property of "Boss", that is the primary key in the data table, and then a new term, called " key Attribute ", I think the key attribute is in pairs appear, mainly is the fact table in the primary key and dimension table of the foreign key, double bond, that for the star model, The dimension tables are linked directly to the key attribute, while the snowflake model, only the partial and key attribute hooks, is weak, so to speak, MSDN.

There are so many "attribute creep", in order to facilitate the management, then to the level of these creeps, "Brigade Platoon" the way out, so there is a term, called " attribute hierarchy ", for example, in the "Calendar Time" hierarchy, the "day" level should be related to the "month" level, "month" The level should be related to the quarter level, and so on.

There are so many "attribute creeps", there is a relationship between the creep and the creep level, and all the creeps are close to the Boss "key Properties", so MSDN says that the attributes in the dimension are always directly or indirectly related to the key attribute. So there is a term called " attribute relationship ".

There are so many "property creep", if the level and attributes of these creeps together, to meet the "hierarchy contains each property with its next direct attribute one to many," called " Natural hierarchy Relationship ", the death of the individual, but it is not difficult to understand.

Continued ing

5. Writeback of Dimensions

The data in the dimension is usually just readable data, in some cases may be enabled for the dimension of the write operation, any member of the dimension can be modified, or there are some limitations, MSDN has detailed additions and deletions to the dimension of the description, when used naturally want to think, here do not repeat.

6. Translation of Dimensions

The translation here is the same as what we mean by translating the English into Chinese, and the client can pass the corresponding label to the SSAS instance based on the region identifier (LCID).

SSAS Series--"02" Multidimensional Data (Dimension objects)