My first Publication Agile-Data-Warehouse-Design-eBook | Page 50

Modeling Business Events
29
Discrete Events
Discrete events are “point-in-time” or short (duration) stories. They typically
represent the atomic-level transactions recorded by operational systems. Example
discrete events include:
Discrete events are
“point-in-time” or
short duration
A customer buys a product in a retail store
A visitor views a web page
An employee makes a phone call
Discrete events are completed either at the moment they occur or shortly thereaf-
ter. By “shortly”, we mean within the ETL refresh cycle of the data warehouse; i.e.,
they have “finished” or reached some end state by the time they are used for BI.
Discrete event stories are generally associated with a single verb (e.g., “buys”,
“views”, “calls”) and a single timestamp. There are exceptions to the one verb, one
timestamp rule, but for an event story to be discrete none of its details must change
over time, otherwise it is evolving.
Discrete event
stories are
“finished”. They
do not change
Evolving Events
Evolving events are longer-running stories (sagas) that can take several days,
weeks, or months to complete. They are typically loaded into a data warehouse
when their stories begin. Example evolving events include:
Evolving events
represent irregular
periods of times.
Their stories may
A customer orders a product online and waits for it to be delivered
A student applies for a place on a university course and is accepted
An employee processes an insurance claim
Evolving events often represent a series of discrete events (chapters if you like) that
BI stakeholders view as milestones of a complex/time-consuming business process.
In Figure 2-1 the arrows that follow each evolving order event mark the shipment,
delivery, and payment milestones that have been reached. Each of the verbs:
“order”, “ship”, “deliver” and “pay” can be modeled as separate discrete events, but
from the stakeholders’ perspective the really important measures of the order
fulfillment process only become visible when these events are combined to produce
a multi-verb evolving event story.
not have “finished”
Multi-verb evolving
events combine the
verbs of discrete
events to
support process
performance
measurement
Timelines (have you noticed how much we like them) are a great way to visualize
evolving events stories and an invaluable tool for modeling milestones and
interesting time intervals (duration measures). Modeling with timelines is covered
in Chapter 8.
Recurring Events
Recurring events are periodic measurement stories that occur at predictable inter-
vals, such as daily, weekly, and monthly (serials). In Figure 2-1 the arrowed line
preceding each recurring event represents the period of time that the event meas-
ures. Example recurring events include:
Recurring events
occur at predictable
intervals