October 29, 2019 Thesis Open Access

A self-learning solution for effective service modelling and portfolio management

Daoud, Rasha

Thesis supervisor(s)

Zhao, Zhiming

 Software companies aim to continuously improve the efficiency of delivering solutions to customers and

reduce their software maintenance cost and maintain sustainability as Software evolves. Service Oriented

Architecture (SOA ) cuts on development costs and efforts by allowing developers to model their software

assets with standardized interfaces and to reuse them in new solutions as building blocks. A Service

Portfolio documents the service assets and their evolution managed by the company, including services

in the pipeline, retired services, and 3rd-party services which are an integral part of the service offerings

to customers. It provides a common information source for developers, sales, customer support and other

roles in the company.

The functionality and the granularity of the entities in the portfolio have a direct impact on the

reusability and sustainability of those assets. When developing various SOA solutions and with the

continuous development and delivery in Agile, design decisions must be always taken to slice and realize

the required scope of functionality into services and components. Modelling existing assets as services

with well-defined granularity can not only significantly leverage them in the design loop, and optimize

the development costs, but also improve the maintenance efficiency of the service portfolio by minimizing

the number of less reusable inventories and variations.

Existing methods focus on identifying and creating services in the design loop, but don’t consider integrating

proper granularity and development cost indicators from the portfolio in the modelling decision

pipeline and the evolution of the portfolio. We thus proposed a decision support approach that uses a

collaborative modelling technique for both legacy/ existing assets and new required business functionalities

and uses variation analysis activities in the design loop to discover assets for reuse. Combining this

approach with the continuous development and delivery makes it work as a self-learning design support

system that continuously aligns available software assets with business needs and helps companies evolve

their service portfolio in the right direction for the market.