Information Technology Reference
In-Depth Information
Chapter 6
Methods of Data and Software Migration
6.1 Acceptance Procedures and Data
In many cases, when implementing new systems, a data base, either entirely new or
from a predecessor application, has to be uploaded. These cases are sometimes
referred to as data bootstrapping. When updating software or after changes to the
data model existing data have to be integrated into the new software version. This is
called migration.
In case of data bootstrapping data from predecessor systems have to be adapted
to the formats of the receiving system. Since quite often these data are of significant
size this is usually done by program. It is therefore recommended to subject the
overall data bootstrapping process to acceptance procedures as well. The same
should be done for migrations. As a consequence more time between software
acceptance and putting into production is required. This has to be taken into account
in the planning process.
The necessary time not only refers to the technically required time to upload a
dump or update a data base. Much more time is necessary to check out the data
quality in the target system, after uploads have been completed. Often data records
have a different structure than in the source system, or new items have been added
or existing ones left out. The total quality is, however, not only measured by the
number of record rejections during migration but has to be checked for content by
random sampling of transferred data.
During preparation already existing data to be transferred have to be cleaned up
before, relieved from “dead wood” and inconsistencies. This work should not be
done later on in the target system.
Of course, as with everything else data bootstrapping or migration is mostly
done under time pressure. This should not discourage the project. Again: these tasks
have to be done one way or another anyway. What will be neglected during the
preparation phase has to be caught up later by cost intensive data clean up projects.
