At 2011's GSE Management Summit, the first GSE Academic Award for Excellence was bestowed in context of the IBM / GSE Academic Initiative. The contributions should address academic work, that GSE and IBM acknowledges as relevant to the future of enterprise computing. As the lucky winner of the Grand Prize, I would like to give you a broad insight into my ideas for improving the world of computing a little bit.
First of all, allow me to introduce myself: I am Fred Stefan, PhD student within the department of Business Information Systems at the University of Leipzig (Germany) . In our research group, we concern ourselves with the integration of business information systems in a typical German engineering-like manner – we call this integration engineering. That concerns everything around the topic of putting different business information systems together. Our main research interest is the “glue” needed to conglutinate these systems together. This leads us right to the underlying problem of this contribution: the integration of different applications.
The task of integration constitutes a large challenge in many areas of computer science. Integration projects are impotent elements of a flexible information technology and essential for flexible business concerns. However, a misapplied integration approach can get very time-consuming, complex and expensive. The aim of my award-winning paper was the presentation of “Lightweight And Minimally-Invasive Concepts For Smarter Integration”.
State of the Art
In spite of highly sophisticated integration suites and a multiplicity of academic process models, integration continues to be difficult. There are many ways to do integration - but there are no firm rules as to why something is done in a certain way. Mostly the line of the least resistance will be chosen. Furthermore, the usage of academic approaches in practice remains unclear. To fill the obvious gap between academic considerations and practice, we conducted a qualitative, empirical study  among 30 integration service providers in the form of expert interviews. Up to now, no comparable attempts are known, considering integration from such a practical perspective. The aim of the study was to examine which integration methods discussed within academia are used in practice and how integration services are provided today. Finally, our study presents on over 270 pages the accumulated knowledge of theses 30 integration experts.
So regarding the title of my contribution, one of your questions might be: Why does integration need to be done in a smarter way? When we asked the experts which factors influence the decision of their customers when they have to decide for a specific solution, the three main arguments were costs, flexibility and time:
- Integration should be cost-effective and get along without further, perhaps wickedly expensive, new hard- or software.
- The aspects of flexibility, changeability and scalability of the integrations solution are important for both development and runtime.
- Integration should be fast. Thus, clients can fully concentrate on their core competencies and due to the short time-to-market, they are able to develop unique selling propositions.
So the way in which integrators conduct an integration project plays an obviously crucial role. The trend is towards lightweight integration approaches.
Likewise, it turned out that mainframe systems and legacy applications represent an own class of integration problems. The experts described for example missing interfaces; monolithic and old-fashioned code structure; limited graphical user interface elements; busy development departments; nebulous responsibilities; lack of skills and expertise or minimal, missing, or out-of-date documentation as common problems that emerge with mainframes and their applications. Since almost all of the mainframe systems represent high client investments and incarnate reliability and quality of service, they should not be touched in many cases. For this reason, the invasiveness of integration is an important point and has to be considered by the integrators. The brisance of this topic does not only exist in the legacy field, but also in all other integration contexts. One of the difficulties of many integration projects is that they often require modifications to existing systems. Moreover, many organizations do not want to change anything in their existing (IT-) infrastructure. While any modification is risky and expensive in terms of development time and skills, it may be desirable to invade the applications, systems and organizations as little as necessary.
As the study shows, the terms of weightiness and invasiveness are broadly known in practice, but there are no satisfying definitions of these two integration aspects. At this point, my contribution attempts to close this gap and presents a first concrete example of how to integrate a legacy application in a minimally-invasive manner.
Due to the limited space of this article, I shall not refer to concrete concepts, but rather I would like to give you a rough overview. Some time ago, the field of software engineering was faced with nearly the same issues, ie discussions about cost reduction, shorter project timelines, less planning effort and enhanced agility. The obvious dichotomy between careful planning and the effort to manage possible changes caused many discussions. Finally, they solved the problems with agile, or also referred to as lightweight, process models.
Similar to software engineering tasks, integration projects are also confronted with comparable challenges, which are manifested by the client’s requirements. However, what defines the weightiness of an integration approach? Generally, the weightiness is determined by the applied approach. That means how the integration project is finally realized.
- Lightweight integration approaches put a special value on short development cycles paired with a minimization of risks and costs. In order to reach this, among other things, the following possibilities exist:
- Iterative and incremental development: An iterative or incremental development gives the integration project the necessary agility to react on arising changes.
- Prototyping: Prototypes allow the integrators to get first integration-technical experiences to reach the integration goal fast and economically. Another point is that first results become very quickly visible, so that the clients can easily try out how a potential solution feels in practice.
- Limited problem space: Furthermore, one should not try to change the whole world in one integration project. If possible, the problem space should be limited to reduce the overall complexity of the integration problem.
If justifiable, the applied integration approach should be as lightweight as possible.
As previously mentioned, the change and invasion of integration objects and organizations is a further important aspect of integration. Generally, it can be distinguished between invasive and non-invasive integration approaches. The fundamental understanding of a non-invasive integration is, that there are no modifications or additions to existing applications or involved systems necessary to integrate them. So that integration for example is only a configuration issue.
But what is minimally-invasive integration? The basic idea of this conceptual design originates from surgery, where minimal invasiveness indicates a low degree of injury during an operation. Brought to the discipline of integration engineering, the main focus is not on humans, but rather on organizational, process and technical levels. The invasiveness of integration is determined by the degree of intrusion and change on these three levels. Invading only the most necessary things saves costs and time and also increases the acceptance of the users. The contribution presented, for each of the three levels, several possibilities and benefits of minimal-invasive integration approaches.
The competition entry was complemented by a small showcase, created in our integration laboratory. To get an insight into this showcase, I would like to explain a little bit more in detail, what invasiveness on a technical level means and why minimally-invasive concepts are very important especially on this level. Imagine: If for example an old Legacy application should be integrated, it could be possible, that the legacy application has no suitable interfaces. In such cases, intrusion is nearly unavoidable. But typically these applications represent substantial investments and cannot be easily changed and rewritten. In these cases, the technical interventions should be as minimal as possible.
To present first results on a concrete example, a mainframe based legacy application was integrated with a modern shop system. A self-developed event- and rule-based observer architecture allows a fast and efficient integration of prevailing traditional and modern application systems, even if no adequate interfaces were available on the side of the legacy application. The integration required only minimal intrusion into the legacy application (in form of database triggers). In addition, the shop system could be integrated in a non-invasive fashion. No additional investments in hardware, software or staff were required at all. For flexible business operations, the rules are easily modifiable and extendable.
The aim of this contribution was to describe lightweight and minimally-invasive concepts for smarter integration. The practical needs of such concepts were deduced by an empirical study.
Naturally, integration projects touch many business and technical aspects within the organizations. However, it turns out, that not only technical challenges influence integration projects. Additional important challenges span far across business and technical issues. Integration projects require a significant slow hand and a gentle touch during the realization. In this context, particular lightweight integration concepts have been presented. Furthermore, these concepts provide the necessary flexibility, to adapt changing business needs and dynamic markets.
Nevertheless, I also focused on invasiveness, as an important attribute of integration projects. I described, on which levels invasiveness occurs and how integrators can deal with it. The benefits of the introduced concepts are closely aligned to the business benefits: the overall objective is to save costs and time. Furthermore, the problem of adequacy of the integration solution is an important point. You do not need a sledgehammer to crack a nut! In summary, the advantages of lightweight and minimally-invasive concepts are: reduced costs for integration, reduced time-to-market for new products and services, early and short result delivery cycles and only as much intrusion as necessary.
Further research has to be done regarding integration metrics. To measure weightiness and invasiveness, metrics have to be developed. For example, reasonable measurement categories are maybe: project budget, (financial) impact, involved staff, involved systems, adoption rates, changed lines of code, operating costs, training course expenditure, etc. Beside measurement categories, reasonable evaluation parameters and target values have to be derived.
Further hot topics
In order to arrange our research efforts as much practical as possible, we are always looking for application partners for corresponding research projects. Looking at the current movement into the cloud, one hot topic is the question, how integration services should be specially provided for these cloud applications and data. Another interesting research question is in accordance with Software as a Service (SaaS) offers the provision of Integration as a Service (IntaaS). In our opinion, these current issues will take on increasing importance in the future.
Importance of the GSE Award
To sum up, it is a great honor for me to be the first Grand Prize winner of the GSE Academic Award for Excellence. In a certain way, this award confirms the importance of our research efforts in the field of integration. However, I was very pleased to be invited to present my ideas in Madrid. In addition, it was a very interesting event coupled with many deep and meaningful discussions. One of the other positive side effects was that I could discover in beautiful autumn weather the city of Madrid.
Events, like this GSE Management Summit build an important bridge between the academia and the practice, on the one hand, by moving the students closer to real life business experiences and on the other hand, giving all participants insights into the highly exciting world of actual science and research topics. Efforts should be made to work constantly on the expansion of this bridge in form of research projects.
PhD student at University of Leipzig
 GEBAUER, M. ; STEFAN, F.: Systemintegration : Eine qualitative Erhebung aus der Sicht von Integrationsdienstleistern. Leipzig : Eigenverlag Leipziger Informatik-Verbund (LIV), 2011 (Leipziger Beiträge zur Informatik Band XXVIII). – ISBN 978-3-941608-15-3 (free download version will be available soon)
GSE Student Award for Excellence sponsored by