Skip to main content

An experiment with an automatic repository based impact analysis system mechanism


Demonstrate the improvement of the software development- and delivery process in a limited experiment directed towards solving the problem of conflicting components within complex information environments, using standard software packages. The expected result will be a significant improvement of the quality of implemented software thus reducing the number of errors occurring during delivery and during execution of software systems. Targets: reduce delivery-errors to zero, reduce errors related to impact by 50%, reduce capacity needed by 60%, Increase availability to 24 hours a day, 7 days a week.

Banks and other financial institutions in general use complex information environments (hardware, software). Changing software-components in these environments is an activity with a high risk, because of the (mostly not completely known) complexity of the relationships between the components under change.
Also other financial institutions are forced to change their information systems because of the year 2000 and the introduction of the Euro. ABN AMRO Bank as a major financial institute uses its experience with complex information environments to perform the experiment and wants to exchange the experience on a European level with other organisations.


The aim of the experiment is to demonstrate the improvement of the software-delivery process using a Repository-based fully automated impact-analysis mechanism.
After the repository has been populated with all components and all relationships between these components, a change to any component will signal the impacted components to their owners and will automatically regenerate the affected components.
To be able to control a large number of projects, often making changes to or affecting the same set of components, a release-control mechanism will be used to order the delivery of the changes.
This release-control mechanism will be included in the repository.

The project team will consist of 9 people during the building-phase.

The experiment will be performed on all projects which will deliver changes to ABN AMRO's MVS production environment during the period September-December 1996. The preliminary results will be used to further improve the system. Some of the projects will make significant changes to the administrative processes related to the most integrated administrations available, the financial and clients administration.


We expect a major improvement of the time-to-market of our software products resulting in an improved time-to-market of banking products. Above that we will improve the control of the change process, the quality and stability of our information systems and our potential to make changes related to the year 2000 and the Euro.
Future improvement plans based on the results of the experiment will be extending the solution to other platforms thus improving our capabilities to make changes in client/server applications in a controlled manner. Because of the general approach taken, also changes in systems software and hardware can be implemented using the same system.
We plan to further develop the described system to obtain planning and actual information about projects executed, thus making it possible to constantly improve the development process.
To be able to start-up projects faster than today, we also plan to add resources with their skills and skills needed by projects, thus making it possible to help project teams in allocating the proper skills on the planned starting dates.

Funding Scheme

ACM - Preparatory, accompanying and support measures


Abn-Amro Bank Nv
Paalbergweg 9-11
1105AG Amsterdam