Information Technology Reference
In-Depth Information
Upstream versus Downstream
Upstream and downstream definitions are sometimes very clear, yet some-
times they are arbitrary. If the system consists of a number of processes
transferring data across each other, as often happens in batch processing,
upstream is more easily identified as the process that happens earlier.
When a number of applications are updating the same database (for
example, a purchase order database), it is not clear whether one can be
considered upstream and the other downstream, even if there is some
logical business-related ordering in time. There are situations where this
clarification is necessary. If one is trying to plug a problem, one should
try to attack it at the most feasible upstream location. This makes sense
because it would have the most impact. If a new data validation must be
applied, it should be put in at the data entry screen, which is the most
upstream for that data.
Because the upstream and downstream metaphor links us to rivers, there
is an expectation that there is some source for even the mightiest rivers,
some glacier or spring in a cave that is most upstream. The river image
also implies that only upstream can affect downstream. Pollution occurring
downstream cannot affect upstream water. A downstream application that
has messed up some data cannot affect an upstream application, but only
itself and others further downstream. How valid is this in our systems?
If we were to take our familiar SDLC (software development life cycle)
and cast it in upstream and downstream terms, this would suggest that
requirements are upstream and design is downstream. However, we know
that software development does not maintain such pristine unidirectional
flows. Design can end up affecting requirements. It is better to view such
models as merely sequential, rather than upstream and downstream.
When two systems are interacting and there is flow of data in both
directions, then it is still beneficial to identify the upstream and down-
stream points in each direction. In such situations, the point to note is
that an application or process is not always upstream or downstream, but
can change its position.
Import versus Export
. These differing directions
have different implications. The repercussions of import are more dam-
aging than those of export. With import, one is bringing something
Import is
taking in,
while export is
sending out
into
the system. If it is data that is imported, then without proper validations
this data can be harmful to one's system. The import and export of data
should not be treated with the same degree of seriousness. Such an error
