The inherent complexity of current software systems has reached a critical level. This affects the general deployment complexity in the form of a huge amount of varying non-functional requirements (NFRs). With regard to flexibility and maintenance, changes to
software systems need to be automated in order to deploy different configurations of a system based on a given set of NFRs.
The paper proposes an Automated Software Engineering (ASE) approach which addresses the automated compile time/deployment time adaptation of NFRs in the context of Web Service middleware. It is shown why Web Service standards and a high-level programming language are advantageous for the NFR optimization of a software system, and how flexible, adaptable Web Service middleware (OpenXL) can be implemented in the industry context using ASE techniques. The paper describes the architecture of OpenXL and depicts in detail how an automated middleware composition process is implemented. Additionally, it is illustrated how a single code source can be used for multiple applications which all address different NFRs. Moreover, a standardized benchmark (TPC-W [1]) is utilized in the measurements section in order to evaluate the OpenXL implementation and to draw meaningful conclusions of the accuracy of ASE. Basically, the proposed ASE approach is generally applicable to the automated NFR-adaptation of software systems.