THANKS: Both answers are very useful here, but I could only choose one. I really appreciate the advice!
our datawarehouse will be used more for workflow reports than traditional analytic reports. Our users care about the "current picture" much more than the story. (although history also matters.) We are a government agency that has no costs or associated calculations. Basically just the number of people in certain places and with a related story.
We use Oracle, and I found a great advantage in using star aggregation as far as possible and would like to rebuild everything so that it resembles the star scheme as closely as possible, which is reasonable for our business goals. Speed in this DW is vital, and a series of tests have already proved to me the scheme of the star scheme.
Our "person" table is key: it contains more than 4 million records and will be the most frequently used source in queries. . This can be seen in the center of the star with several dimensions (e.g. age, gender, gender, location, etc.). This is a very long table, especially when I join it at the address and contact information.
However, this is more like a measurement table when we start looking at history. For example, there are two history tables in which there is a person’s key pointing to a table of persons. One has over 20 million records, while the other has nearly 50 million and is growing daily.
Is this table a fact table or a measurement table? Is it possible to work on both? If so, will this be a big performance issue? Is it customary to request more dimensions than fact? What happens if the fact table DIFFERENT, using the Person table as a dimension, is actually only 60,000 records (much less).
I think that my problem is that our data and their use do not correspond to the widely used examples of stellar schemes.
: , , , , . :
. - , : , , ; , .. " ", ( , ..). , , . , .
"-": ( ) . -, - . -, - , , . ( , - - , . , . LOL)
, .