Quite simply, a table such as this:
Dim1ID Dim2ID MeasureID Value --------- ----------- ------------- ------------ 543 44 1 234.3 543 45 1 256.3 544 44 1 245.3 544 45 1 264.5 543 44 2 10 543 45 2 8 544 44 2 9 544 45 2 10
With one value column that signifies different measures using a foreign key.
It is possible to reputation for this pattern?
Entity-Attribute-Value Model possibly?
Editorial: Many people think about this to become an anti-pattern (in SQL), though in Column Based stores, this is actually the Usual Behavior (BigTable, Cassandra).
Appears in my experience that, as
MeasureID would reference a table listing all possible measures, you've just got a star scheam with 3d where one dimension is known as "Measure".
I'd just refer to it as a 4D table: three key characteristics and something non-key. I do not think it requires a unique title.
I labored having a nearly identical model a while ago. We'd up to 8000 measures and many billion rows. Within the DBMS i was utilizing it was totally not practical (and unnecessary) to produce tables with a large number of posts. The "wide" row version might have didn't have values whatsoever for the majority of the measures of all of their rows. Therefore we would either have needed to generate nulls or dummy values for data where none been around or we'd have needed to create 100s of "narrower" tables with nearly arbitrary teams of measures inside them. The "vertical" model makes much more sense and can play very well using the right DBMS.
I disagree using the suggestion that the design isn't correctly stabilized. As lengthy as all of the measures are of the identical data type then its the best design and it is in a minimum of fifth Normal Form (presuming the dimension and measure identifiers make up the key). An alternate design with a lot of posts certainly would not be stabilized whether it forced you to employ nulls.