Benefits of AssessGrid approach

AssessGrid outcomes can be applied in various sectors with high compute needs that embrace the Utility Computing model, for instance, financial sectors, telecommunications, automotive, and energy sectors.

The provisioning of services with an agreed Quality of Service (QoS), risk assessment and management, and optimisation of the use of resources is of great benefit to resource providers. The key drivers for Grid commercial uptake are increased efficiency and cost reduction. Both of these areas are well addressed in AssessGrid, with its emphasis on risk assessment models and capacity planning for QoS provision. At present in many of the resource providers’ activities there is a management gap with too much reliance placed on expert intervention to actually tune Grids. For Grid computing to be adopted in a commercial space, it is essential that many of the resource management tasks are undertaken in as automated as fashion as possible, including risk assessment.

For a resource provider such as Amazon, Google, or BT the use of the AssessGrid risk models means that Service Level Agreements requests can be assessed carefully, and time critical operations can be planned effectively. For example the downtime associated with a critical situation can be minimised, major commercial assets (e.g. equipment undergoing maintenance) can be utilised efficiently and expensive staff can be deployed more usefully. A large number of Grid applications, through SLAs, will benefit from the risk assessment models, which include for example engineering services and maintenance planning, commissioning and optimisation of equipment, financial forecasting etc. Using more efficiently Grid resources is a gain in the investment. This will not only ensure cost savings for the resource providers, but clear benefits for the end-users as well. 

In terms of cost (which we foresee as small) some technical settings are certainly required for resource providers to make use of AssessGrid software prototype. The AssessGrid components such as the Negotiation Manager which enable the communication between the broker and the end-user portal need to be integrated.

The AssessGrid consortium will deploy its software prototype on real Grid resource provider infrastructures to carry on an effective qualitative evaluation of the benefit of risk assessment. Basically this will involve setting a set of experiments to compare the performance of the Grid infrastructure with and without AssessGrid components, particularly the risk assessment ones. Plans are under way with the UK National Grid service (NGS), which has shown strong interest in the project outcomes. The scenario is to deploy the broker and the resource provider software components on the NGS and measure their effectiveness from the perspective of the three main actors: resource provider, broker, and end-user.

The broker’s risk-awareness means the broker service is used by the end-user in order to find suitable providers for the execution of a given job (described as a SLA request), pre-negotiate a SLA with different candidate providers on behalf of the end-user and act as a virtual provider, agreeing its own SLAs with end-users. The benefits of using the broker service can be measured in terms of metrics such as number of SLAs fulfilled, effectiveness of fault-tolerance mechanisms, and response time.

End-users will express the need for the use of the AssessGrid broker service if they were unsatisfied with QoS received before its deployment on the NGS, e.g. no fault-tolerance mechanisms in place following an SLA failure, unreliable resource provider etc. In order to guarantee a more efficient service, the end-user will be able to evaluate resource providers through the AssessGrid Broker Service to check their reliability and the evaluation of risk the end-user incurs with each of the NGS resource providers. The benefits for the end-user can be measured in terms of metrics such as number of SLA objectives fulfilled, flexibility, and response time.

The project has also performed a quantitative analysis of the benefits of probabilistic and possibilistic Risk Assessment methods which is available on a dedicated page.