High-Quality Business Processes Based on Multi-Dimensional QoS

High-Quality Business Processes Based on Multi-Dimensional QoS

Qianhui Liang (HP Labs, Singapore) and Michael Parkin (Tilburg University, The Netherlands)
DOI: 10.4018/978-1-60960-794-4.ch007
OnDemand PDF Download:
No Current Special Offers


An important area of services research gathering momentum is the ability to take a generic business process and instantiate it by selecting services that meet both the functional and non-functional requirements of the process owner. These non-functional or quality-of-service (QoS) requirements may describe essential performance and dependability requirements and apply across different logical layers of the application, from business-related details to system infrastructure; i.e., they are cross-cutting and considered multidimensional. Configuring an abstract business process with the “best” services to meet the process owner’s multidimensional end-to-end QoS requirements is a challenging task as there may be many services that match to the functional requirements but provide differentiated QoS characteristics. In this chapter we explore an approach to discover services, differentiated by their QoS attributes, to configure an abstract business process by selecting an optimal configuration of the “best” QoS combinations. The approach considered takes into account the optimal choice of multi-dimensional QoS variables. We present and compare two solutions based on heuristic algorithms to illustrate how this approach would work practically.
Chapter Preview


When a business analyst designs a new business process or performs refactoring of an existing process, determining the dependability and reliability of the final end-to-end composition of operations and services that make up the process is of paramount importance; without considering aspects of the business process such as the maximum throughput, time required to complete the process or its availability, the process might not meet the expectations of the business or their customers. Therefore, an abstract business process, which we consider to be a collection of interrelated, structured activities or tasks, must be designed not only with coherent and functionally compatible services but also with a set of qualities of service (QoS) that will fulfill the necessary business objectives.

This chapter focuses on the dependability of business processes by considering how an abstract business processes is deployed with a selection of services from the Internet of Services that meets the desired end-to-end quality of service (Cardoso et al, 2009) from the process’ system-level perspective (Liang et al, 2008) (Liang et al, 2009). The challenge with operating a business process with end-to-end quality of service like this is that no methodology exists for determining resource allocation from the many autonomous service providers available through the Internet of Services.

This chapter demonstrates how an approach based on a multi-criteria analysis can be used to decide on the combination of services to select. Later, we describe how we implemented this decision making technique using two optimization techniques (simulated annealing and genetic algorithms) and carry out an experimental evaluation of the performance of each technique. However, we first present an overview of the context and issues involved and describe the proposed high-level approach before introducing the decision making techniques.

Complete Chapter List

Search this Book: