any idea, if (and if yes then how) small datatypes as TINYINT (= 1 byte) could be compressed in columnstored indexes?

Its clear, that I will save space when I replace "this is a long string" with a simple tinyint (at least, when it occurse more than once), but when the column data type is already as short as possible, it could only work when you save "value 1 (which is the dictonary entry for the original tinyint CountryID 20) will used in rows 1-10, 12-50 and 90-200", but this would it make much harder to recombine rows to answer an SELECT *