Information Technology Reference
In-Depth Information
functional specifications. Furthermore these specialists quite often have direct
access to the responsible developers. This facilitates rapid clarification.
The patch cycle agreed upon prior to acceptance tests is particularly helpful
allowing practically immediate error resolution (with only minor temporary incon-
venience for the test business) by hot fixes for example. This can be followed up
directly by additional tests of the corrected versions.
Software corrections resulting from production problems at a later stage have to
follow the scheduling of updates. A prompt implementation therefore is only
possible in exceptional cases—for example in extreme breakdown situations—
with all the risks for active production during uploading and later executing
untested changes. When changes to the data model have to be done, longer
downtimes are unavoidable.
3.2.4 Avoidance of Data Inconsistencies
When linking different systems, each of which cover specific parts of a business,
and which exchange data and have access to the same data base, the problem of data
consistency may arise. For systems having been designed from scratch this problem
should be negligible or non-existing. However, it may become virulent once one or
more existing systems are linked up, since every system including its associated
data pool has its own history.
Later, when dealing with data migration (Chap.
6
) preventive measures will be
discussed. Indispensable, however, is a carefully prepared interface test, if possible
with real life data, not only to ensure interface functioning but also to check data
quality regarding consistency and rejection rates. Data clean-up projects to be
initiated later for production data files, because previous testing had been neglected,
are very costly and may take a long time. In most cases those measures will lag
behind the ever changing actual data base constantly modified during production. It
can be very costly with regard to data that are visible to the outside world like bills
to customers for example.
To recapitulate:
The economic aim of quality management during implementation for com-
plex IT systems is the avoidance of heavy costs during later stages. Experi-
ence shows that omission of investments at the early project stage quite often
results in later amendments, costing up to 200 % or more of preventive
measures.
