Thomas Unterschütz, Veit Vogel

T-Nova Deutsche Telekom Innovationsgesellschaft mbH
Technologiezentrum Darmstadt
Am Kavalleriesand 3, 64307 Darmstadt
phone: +49.6151.83-1288, fax: +49.6151.83-4590

Growing complexity and size of nowadays Software systems, in business as well as in front end applications require sophisticated tools and methods for abstraction from underlying operating systems, hardware and location. The introduction of distributed object technologies is the most important step to ease the development and management of huge applications. Distributed object oriented platforms enable an integrated approach to software engineering, including issues such as heterogeneity, interoperability, security, data storage, -access and -management , transport networks transparency and many more services. Moreover for the application developer the DOT products offer a wide range of basic services with no need of thorough knowledge of the particular service implementation.

The availability of QoS-Aspects in DOT is still left undone. Yet aspects like real-time guaranties, resource availability and selection of underlying transport network QoS are not supported and therefore are not available to the application.

The proposed presentation introduces ongoing work of a European project on solving distribution and configuration support for DOT. Methods and tools are to be developed to facilitate and automate the deployment of large object oriented, component based applications on CORBA-platforms, and to achieve best performance of the applications and efficiency of platform resources use. A specific focus of this work is influenced by the telecommunication field, due to the business and technical background of the project partners. The information presented is based on the work done in the EURESCOM project P924 "Distribution and Configuration Support for Distributed PNO Applications".

The presentation will include the following substance:

Static distribution and configuration of components for complex applications to a platform require thorough knowledge of the components’ requirements concerning the platform topology and the nodes’ environment. To automate the distribution of components on a given topology of nodes, a formal notation of distribution parameters is need as well as information on the configuration of the platform, i.e. capabilities of nodes, on which the application is going to be deployed. An optimised component distribution pattern will be automatically calculated matching the components requirements with the platform nodes’ capabilities. Furthermore automatic distribution and configuration tools will be developed in order to support distribution and configuration of the components to the platform according to the calculated distribution pattern.

Moreover looking to the dynamic aspects of distributed, component based applications, continuous monitoring of the applications’ distribution parameters is needed to ensure they match the contracted Quality of Service, the applications users expect. Due to the dynamic nature of distributed applications and the abstracting quality of distribution platforms, many unforeseeable events might cause changes of the applications’ QoS, which must lead to redistribution (migration), reconfiguration or duplication (scalability) of the applications’ components. Reconfiguration and extension of the platform itself is another option for solving tight QoS corners. Especially considering complex telecommunication’s applications, globally distributed on huge platforms, it is of high importance to automate these tasks, and to not bother the applications’ operators with the manual recalculation of distribution patterns and dynamic reconfiguration of the components.

The work presented is related to the introduction of distributed component based telecommunication services. Yet these new services are mainly deployed on small and medium size distribution platforms, but the growing demand for complex and individually configurable services is already foreseeable. This induces the need for larger platforms and a more dynamical handling of versatile and reusable service components to gain best performance and best use of resources.