etl - Which is the best approach for a dimension (SCD-2 or a SCD-1 + a whole new dimension) -
let´s have following situation:
a dimension product attributes aren't volatile (description , diameter - can changed scd-1 change correction) , attribute can volatile (selling group, can change on time same product).
so, when change occurs in these volatile attributes of 1 product, need somehow track them.
i have come these 2 approaches:
for both: keep using scd-1 non-volatile attributes.
approach #1: use scd-2 in product_dim volatile attributes.
approach #2: make selling group whole new dimension , every sell track current value in moment of etl. no need scd-2 here.
i new in data warehousing , i'm trying understand better , why. 1 of aims use olap software read of stuff.
it comes business needs of model. don't know business enough question, rule of thumb if wanna analysis selling group (i.e: total quantity of products sold selling group x) should create separate dimension. in case approach#2 correct. considering general concepts , assuming selling group kind of group of products, doesn't make sense having attribute of product.
if want learn more dimensional modelling i'd suggest looking ralph kimball's work if haven't done yet. excellent resource book the data warehouse toolkit covers question , many more techniques. it's nice tool have on desk when questions pop up. of experienced data modellers have copy of consult every , then.
Comments
Post a Comment