Quantcast
Channel: Debunking Kimball Effective Dates
Viewing all articles
Browse latest Browse all 83

re: Debunking Kimball Effective Dates

$
0
0

p.s. I should also add that in Peoplesoft, in addition to effective date, there is of "effective sequence."  In other words multiple rows with the same effective date (no time component), ordered by that seq.  For purposes of dimension data range queries ("what did Joe Blow look like this day"), you want to return the data in the highest sequence number record for that effective date.  But you also need to keep the intermediate seq nbrs because some transactional events (facts) may have happened that could be tied explicitly to the intermediate state (aka not the highest seq nbr for the date).  In my approach of precomputing and storing, I can create a variation of the code you show to make the end effective date of those intermediate rows the day BEFORE the begin date, so they automatically disappear from range queries, and my users don't have to deal with the extra headaches of sequence number sub-ordering. This is not something I feel I, as a data professional, should leave for my end users to deal with. Like I said, I've seen others do that, and I've heard the frustration from the users.


Viewing all articles
Browse latest Browse all 83

Trending Articles