Trusted Grid Workflows: Assigning Reputation to Grid Service Providers
by Omer Rana, Luc Moreau and Julian Padget
Three Projects (PASOA, GENSS, SERENITI) extend the capability of workflow systems in Grid environments generally focusing on the capacity to enable more reliable, trustworthy and efficient use of Grid resources. Such issues are also significant to encourage greater use of Grid infrastructure by commercial/business users.
Creating applications by combining independently-managed (often distributed) services forms a key theme in Grid computing. Such services may be hosted by different institutions, and utilise a variety of compute and data resources. Workflow tools play an important role in helping compose these services, and considerable effort has been put into creating workflow tools that combine ease-of-use with support for the widest range of Grid middleware. Most Grid-centric workflow tools assume the existence of a static set of services, available at the time of launching the tool. To exploit fully the dynamic aspects of Grid infrastructure, it is necessary to allow for dynamic services to be deployed and discovered.
In automated discovery techniques, services have a rich semantic description and a matchmaking algorithm attempts to match the service request with service advertisements. The match is then classified as an exact match, or several flavours of relaxed match. It is important to note that often automated discovery also requires having a human in the loop as it is seldom possible to specify user needs in an exact way (often users are also not fully aware of their requirements). It is intended that the identification of sufficiently similar services to a query is often likely to lead to a discovery of more suitable services. The Grid Enabled Numerical and Symbolic Services (GENSS) project is an EPSRC-funded project between the University of Bath and the Welsh eScience Centre (WeSC) at Cardiff University. The project builds on the development of Mathematical Web Services in the European Commission funded MONET (http://monet.nag.co.uk/) project. The GENSS project addresses the creation of matchmaking techniques to allow users to discover Mathematical services described in terms of an OpenMath-based ontology. Mechanisms to allow such services to be delivered as part of a workflow for mathematical problem analysis is also being undertaken in the project. Such service discovery, by necessity, needs to be problem-specific. By focusing on mathematical services, we intend to address the requirements of a much wider range of users in science and engineering than perhaps by focusing on a particular application domain (such as BioSciences, for instance).
Once suitable services have been discovered, it is often difficult for the user to ascertain if the service selected will perform as expected. This introduces the issues of trust and reputation in services made available over Grid infrastructure. Trust issues may be considered from the service provider's or the service consumer's perspective. For the provider, trust has always been associated with security -- focusing on authenticating and authorising users accessing the provider's resources. For the consumer, trust is considered as a predictability issue addressing concerns such as which service provider is truly reliable in delivering a service, and how much credence should be given to that provider to deliver the service as advertised. The concepts of trust and reputation are complex and many-faceted issues and associated with themes such as risk, competence, beliefs/perceptions, utility/benefit and expertise. Two projects at WeSC are attempting to address these concerns:
- PASOA (Provenance Aware Service Oriented Architecture), led by the University of Southampton (EPSRC-funded)
- SERENITI (SERvice-ENabled Interactive Trust Informatics), led by WeSC.
The PASOA project is investigating how provenance data may be associated with a workflow process (process provenance) and with each service (service provenance). The problems of determining the origin of a result, especially when it involves processing through a number of stages, or deciding when results of analysis are no longer valid become important concerns in open Grid environments and in establishing the quality and reproducibility of scientific applications. The PASOA project therefore is aiming to define execution and service provenance in relation to workflow enactment, to enable better use of this data for reasoning about the processes involved in a scientific collaboration over the Grid.
The SERENITI project on the other hand is exploring how the concepts of trust and reputation are related to each other, and especially how a reputation index may be assigned to a group of services, based on the reputation of each service within the group. The key theme in the project is the observation that in a workflow session there is often a set of critical services which must perform successfully. These critical services are also only available from a restricted set of providers (such as a particular numeric solver that can only execute on a restricted set of platforms). Allocating a reputation-index to such critical services, and also determining how this value changes over time is a key factor in assessing risk for the entire workflow.
Omer F. Rana, Cardiff University, Wales, UK
Tel: +44 29 2087 5542
Julian Padget, University of Bath,
Luc Moreau, University of Southampton, UK