statement:Original Works，when reproduced, please indicate the article from
The SAP division is tooTechnology Blog ( Bo/guest/Park www.cnblogs.com): Www.cnblogs.com/jiangzhengjun，and indicate the original source of the article in the form of a hyperlink，Otherwise the legal liability will be investigated! Original link: http://www.cnblogs.com/jiangzhengjun/p/4297831.html
- Try not to put too much detail in the cube (that is, the smaller the dimension field, the better), the need to first consider r3 with abap resolution, if you want to bw , you can consider dso out of detail report, in the Span lang= "en-us" >cube out of the summary report and calls the detail report through the rri interface.
- split multiple: cube is large, it can be split into multiple cube, and together so that query will n cube in parallel to improve efficiency. This is called a logical partition. Common partitioning method has by date, by country, press bu , by type, etc.
- compression: Give cube do compression . compression essentially removes data Dimension so fact The table is compressed, but request ID also disappears and will not be able to manage data through request ID Span style= "mso-bidi-font-weight:bold;" > caution, preferably half a year or more of data) .
- index: Database index can speed up query speed
- partition: For a large cube , you can do partition, which is a physical partition and only supports partitioning by time.
- aggregation: Using aggregation can improve performance. But aggregation itself is a subset of cube , improving performance while also increasing data redundancy, so don't use too much.
- staitics : regular refresh db Statistics can improve the efficiency of reporting .
- use mp : dimension design, avoid a lot of data size Span lang= "en-us" >char. On a dimension (many-to-many do not put in one dimension), because this makes the dimension table very large. In general, as much as possible to split into more dimensions (but too many dimensions will cause fact table Huge, so do balance), and then multiprovider level, the relevant (refers to the business sense of relevance) char all put a dimension, and then do a mapping , This makes it easier for users to understand multiprovider , Lower cube dimensions from a technical perspective split and combined (by 1:1 or 1:n to improve performance, both worlds.
- line item Dimension : for material , using line Item Dimension (m : n Many-to-many fields, let them one by one dimension, so naturally become the line item dimension, improve performance) .
- BIA: using BIA is a much more effective method than Aggregation , which is to spend a lot of money.
Optimization of InfoCube Information cube
