12,000

We have over 12,000 students, from over 100 countries, within one of the safest campuses in the UK

93%

93% of Lancaster students go into work or further study within six months of graduating

Home > Research > Publications & Outputs > Supporting dependable distributed applications ...
View graph of relations

« Back

Supporting dependable distributed applications through a component-oriented middleware-based group service

Research output: Contribution in Book/Report/ProceedingsChapter

Published

Publication date2004
Host publicationArchitecting Dependable Systems II
EditorsRogério de Lemos, Cristina Gacek , Alexander Romanovsky
Place of publicationBerlin
PublisherSpringer Verlag
Pages270-293
Number of pages24
ISBN (Print)978-3-540-23168-4
Original languageEnglish

Conference

ConferenceWorkshop on Algorithms and Data Structures
CountryCanada
CityOttawa
Period30/07/031/08/03

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume3069
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceWorkshop on Algorithms and Data Structures
CountryCanada
CityOttawa
Period30/07/031/08/03

Abstract

Dependable distributed applications require flexible infrastructure support for controlled redundancy, replication, and recovery of components and services. However, most group-based middleware platforms, which axe increasingly being used as implementation environments for such systems, fail to provide adequate flexibility to meet diverse application requirements. This paper presents a group-based middleware platform that aims at maximal flexibility. In particular, flexibility is provided at design time, deployment time and run-time. At design and deployment time, the developer can configure a system by assembling softwaxe components shaped to a specific use. Then, at run-time, s/he can dynamically reconfigure the resulting system to adjust it to new circumstances, or can add arbitrary machinery to enable the system to perform self-adaptation. As examples, levels of fault tolerance can be dynamically increased and decreased as desired by adding, removing or replacing replicas; or the underlying communications topology can be adapted by switching from point-to-point TCP to multicast as numbers of replicants increase. Importantly, it is not necessary that the shape that run-time reconfiguration takes has been foreseen at design or deployment time. Our proposed solution employs softwaxe component technology and computational reflection as the basic means by which to perform and manage configuration and reconfiguration.