Databases Reference
In-Depth Information
contingencies are built-in and sound risk management procedures are fol-
lowed. There may be a number of static tables, perhaps used for code look-
up that can be converted without as much fanfare, but the main conversion
will take time.
At the time planned for cutover, the old production system can be frozen
from update or run in parallel. The production database then can be initial-
ized and test records removed (if any have been created). The conversion
and any verification and validation routines can be run at this point.
Checking Audit Reports
Once the conversion is finished, special audit reports should be run to
prove that it worked, to check control totals, and to deal with any prob-
lems. It may be necessary to roll back to the old system if problems are ex-
cessive. The new application should not be used until it is verified that the
conversion was correct, or a lot of work could be lost.
Institutionalizing
In many cases, as in data warehousing, conversion will be a continuous
process and must be institutionalized. Procedural controls are necessary
to make sure that the conversion runs on schedule, results are checked rig-
orously, rejected data is dealt with appropriately, and failed runs are han-
dled correctly.
DATA QUALITY
A strategy to identify data problems early in the project should be in
place, though details will change according to the project. A preliminary in-
vestigation can be done as soon as the old physical data model has been
determined. It is important to document the quality of the current data, but
this step may require programming resources. Customers at all levels
should be notified if there are data-quality issues to be resolved. Knowl-
edge of the extent of data-quality problems may influence the user's deci-
sion to convert or abandon the data.
Keeping the Data Clean
If the data is corrected on a one-time basis, it is important to ensure that
more erroneous data is not being generated by some faulty process or pro-
gramming. There may be a considerable time interval between data correc-
tion and conversion to the new system.
Types of Data Abnormalities
There may be integrity problems in the old system. For example, there
may be no unique primary key for some of the old files, which almost
Search WWH ::
Custom Search