Designing fault-tolerant SOA based on design diversity

Amanda S Nascimento,Patrick Hs Brito,Rachel Burrows,Cecília Mf Rubira,Fernando Castor

doi:10.1186/s40411-014-0013-7

Abstract

Over recent years, software developers have been evaluating the benefits of both Service-Oriented Architecture (SOA) and software fault tolerance techniques based on design diversity. This is achieved by creating fault-tolerant composite services that leverage functionally-equivalent services. Three major design issues need to be considered while building software fault-tolerant architectures based on design diversity: (i) selection of variants; (ii) selection of an adjudication algorithm to choose one of the results; and (iii) execution of variants. In addition, applications based on SOA need to function effectively in a dynamic environment where it is necessary to postpone decisions until runtime. In this scenario, control is highly distributed and involves conflicting user requirements. We aim to support the software architect in the design of fault-tolerant compositions. Leveraging a taxonomy for fault-tolerant systems, this paper proposes guidelines to aid software architects in making key design decisions. The taxonomy is used as the basis for defining a set of guidelines to support the architect in making decisions related to fault tolerance in SOA. The same taxonomy is used in a systematic literature review of solutions for fault-tolerant composite services. The review investigates how existing approaches for fault-tolerant composite services address design diversity issues and also specific issues related to SOA. The contribution of this work is twofold: (i) a set of guidelines for supporting the design of fault-tolerant SOA, based on a taxonomy for fault tolerance techniques; and (ii) a systematic literature review of existing solutions for designing fault-tolerant compositions using design diversity. Although existing solutions have made significant contributions to the development of fault-tolerant SOAs, there is a lack of approaches for fault-tolerant service composition that support strategies with diverse quality requirements and encompassing sophisticated context-aware capabilities. This paper discusses which design issues have been addressed by existing diversity-based approaches for fault-tolerant composite services. Finally, practical issues and difficulties are summarized and directions for future work are suggested.

Highlights

Nowadays, society is highly dependent on systems utilizing Service-Oriented Architectures (SOA) for its basic day-to-day functioning (Huhns and Singh 2005; Papazoglou et al 2007)
The adoption of software fault tolerance techniques based on design diversity has been advocated as a means of coping with residual software design faults in operational software (Lee and Anderson 1990)
To mitigate the risks of employing an inadequate taxonomy, before it was built, we had analysed the domain knowledge of software fault tolerance techniques based on design diversity in depth (e.g. (Elmendorf 1972; Horning et al 1974; Kim 1984; Laprie et al 1990; Lee and Anderson 1990; Lyu 1996; Pullum 2001; Scott et al 1987))

Summary

Introduction

Society is highly dependent on systems utilizing Service-Oriented Architectures (SOA) for its basic day-to-day functioning (Huhns and Singh 2005; Papazoglou et al 2007). The use of techniques based on design diversity to tolerate software faults was widely criticised since variant software components used to be developed from scratch, which is very expensive (Anderson et al 1985; Vouk et al 1993). These techniques were generally used only in highly critical systems, in which the occurrence of failures would result in large financial losses or even loss of life. Software developers have been evaluating the benefits of both Service-Oriented Architecture (SOA) and software fault tolerance techniques based on design diversity This is achieved by creating fault-tolerant composite services that leverage functionally-equivalent services. Multiple services running over heterogeneous systems may interact and be used as building blocks for new applications (Papazoglou et al 2007)

Objectives

Methods

Results

Discussion

Conclusion