Databases Reference
In-Depth Information
Asserted Versioning temporal model
that recognizes future
assertion time.
But the hard work has to be paid for. And sometimes the
results are flawed. Sometimes we don't pull together all the data
relevant to a business request, because we overlook some sources
we should have known about. Sometimes we don't pull together
the best or most current copy of relevant data because we don't
know how to tell the best from the rest. And often, by the time
the hard work is over and the data is assembled, that data has lost
some business value, or perhaps has become completely irrele-
vant, simply because too much time has gone by.
When data requests are for our best data about what things
are currently like, we are usually able to respond rapidly. Usually,
the data is already available, ready to be queried from within
application programs, or by means of tools available to business
analysts or researchers. This is the data we have been calling
conventional data. In the terminology we have developed, con-
ventional data consists of currently asserted current versions of
the persistent objects we are interested in.
But past, present or future assertions about past, present or
future versions give us nine temporal categories of data, and
conventional data is only one of those nine. As requests for data
in one or more of those other eight categories become more fre-
quent, it becomes more important to reduce the work involved
in satisfying those requests. As it becomes increasingly impor-
tant to provide that data as quickly as possible, scavenger hunts
for scattered data are less and less acceptable as a standard
procedure.
This increasing importance of past and future data points to
an obvious end-state. First, all nine bi-temporal categories of
data should be as quickly and easily available as conventional
data already is. Secondly, seamless access across data from mul-
tiple categories should be just as easy, and just as quick.
Asserted version tables are that end-state. Asserted version
tables can contain any number of rows representing the same
object, and those tables can include rows from any or all of those
nine bi-temporal categories. This means that when data about
one or more objects, in any of those categories, is requested,
all of that data is immediately available and ready to be queried.
There is no assembly and/or transformation work to be done to
get that data ready for querying. There is no delay between the
business request for data, and its availability in queryable form.
The majority of queries against bi-temporal data are point-in-
time queries. A point in assertion time, and a point in effective
time, are specified. As for assertion time, when Now()
is
