Data, Processes, and Events: Keys to Comprehensive Analysis - High-Performance Web Databases: Design, Development, and Deployment - page 55

Exhibit 4-8. Matrix for identifying events in a pair of entities.
Next considered is the update row of the causal entity. Is there any case
where the update of the ORDER would cause the system to create a LINE
ITEM? Once again, probably not. Is there any case where the update of
ORDER would cause the system to update a line item? This would be a pos-
sibility if the ORDER contained an indication of customer type or order
type that perhaps affects the pricing of each line item. If this is the case, it
would once again be recorded as an event requiring further analysis. Is
there any case where the update of an ORDER would cause the system to
delete a LINE ITEM? Once again, probably not.
Proceeding to the final row (i.e., delete events), the practitioner asks the
same set of questions. Is there any case where the deletion of an ORDER
would cause the system to create a LINE ITEM? Probably not. Is there any
case where the deletion of an ORDER would cause the system to update a
LINE ITEM? Once again, probably not. Is there any case where the deletion
of an ORDER would cause the system to delete an LINE ITEM? Definitely
yes. When the system deletes an order it should delete all the line items
attached to it. Once again, the practitioner records this as another event
requiring further decomposition.
At least two and possibly three major events requiring further explora-
tion have been identified. Exhibit 8 shows how an analyst would illustrate
this in the matrix. Because all relationships are bidirectional, the practitio-
ner must look at causes and effects in both directions.
Looking from the Other Direction
Now the practitioner must go through the same procedure using the
LINE ITEM as the causal entity and ORDER as the affected entity. There
would be no case where the creation of a LINE ITEM causes the system to