Databases Reference
In-Depth Information
Time Variance. The events and objects of the enterprise both share
moments of time. In a d ata warehouse they can be joined together by
that moment in time. A data warehouse user can ask, “Which objects,
attributes, and properties were in effect at the moment of a transaction?”
By joining a transaction to the enterprise objects, attributes, and prop-
erties, a data warehouse with Time Variance can show that transaction
in its enterprise context at the moment of the transaction, which is the
answer to the question above. By joining a large number of transac-
tions, all transactions that occurred during the first quarter of 2011, to
the enterprise objects, attributes, and properties, a data warehouse with
Time Variance can show each of those transactions in their own enter-
prise context at the moment they occurred and yet return all that data in
a single result set.
yeAr, QuArter, moNth, Week, cycle, DAy of
Week, tIme ZoNe, DAte, AND tIme of DAy
Before moving on to the next step we need to inventory what we've learned
thus far about the enterprise, time, periodicity, and how these are all
shared by transactions and objects within the enterprise. First, time is a
concrete abstraction. We cannot touch it, yet we measure it. We cannot
control it, yet we let it control us. Second, an enterprise has many, many
clocks within the enterprise, and each of those clocks performs a function
that is intended to meet the needs of a specific segment of the enterprise.
An enterprise data warehouse must be able to reconcile all the various
forms of time in the enterprise into a cohesive representation of time in
the data warehouse. In doing so, the data warehouse includes transac-
tions that are measured at a fine grain of time because those transactions
happen so many times during a d ay; however, the data warehouse also
includes objects (such as product, price, promotion, and placement) that
change much less frequently, maybe as seldom as once every few years or
as often as once a day. Finally, we can synchronize the state of objects in
the enterprise with the transactions of the enterprise by joining those two
together based on time. In the example we followed earlier in this chapter,
we were able to answer the question “What sales transactions occurred
when the price of gasoline was XYZ?” The answer to that question hinged
on the word
when
.
