Load-balancing for mesh-based applications on heterogeneous cluster computers

This paper discusses load-balancing issues when using heterogeneous cluster computers. There is a growing trend towards the use of commodity microprocessor clusters. Although today's microprocessors have reached a theoretical peak performance in the range of one GFLOPS/s, heterogeneous clusters of commodity processors are amongst the most challenging parallel systems to programme efficiently. We will outline an approach for optimising the performance of parallel mesh-based applications for heterogeneous cluster computers and present case studies with the GeoFEM code. The focus is on application cost monitoring and load balancing using the DRAMA library.

Load sharing in cluster service provision

The factors that are driving the development and use of grids and grid computing, such as size, dynamic features, distribution and heterogeneity, are also pushing to the forefront service quality issues. These include performance, reliability and security. Although grid middleware can address some of these issues on a wider scale, it has also become imperative to ensure adequate service provision at local level. Load sharing in clusters can contribute to the provision of a high quality service, by exploiting both static and dynamic information. This paper is concerned with the presentation of a load sharing scheme, that can satisfy grid computing requirements. It follows a proactive, non preemptive and distributed approach. Load information is gathered continuously before it is needed, and a task is allocated to the most appropriate node for execution. Performance and reliability are enhanced by the decentralised nature of the scheme and the symmetric roles of the nodes. In addition, the scheme exhibits transparency characteristics that facilitate integration with the grid.

Engineering emergence for cluster configuration

Distributed applications are being deployed on ever-increasing scale and with ever-increasing functionality. Due to the accompanying increase in behavioural complexity, self-management abilities, such as self-healing, have become core requirements. A key challenge is the smooth embedding of such functionality into our systems. Natural distributed systems such as ant colonies have evolved highly efficient behaviour. These emergent systems achieve high scalability through the use of low complexity communication strategies and are highly robust through large-scale replication of simple, anonymous entities. Ways to engineer this fundamentally non-deterministic behaviour for use in distributed applications are being explored. An emergent, dynamic, cluster management scheme, which forms part of a hierarchical resource management architecture, is presented. Natural biological systems, which embed self-healing behaviour at several levels, have influenced the architecture. The resulting system is a simple, lightweight and highly robust platform on which cluster-based autonomic applications can be deployed.

An evaluation of the implementation of hand held health records with adults with learning disabilities: a cluster randomized controlled trial

Background: Personal health records were implemented with adults with learning disabilities (AWLD) to try to improve their health-care. Materials and Method: Forty GP practices were randomized to the Personal Health Profile (PHP) implementation or control group. Two hundred and one AWLD were interviewed at baseline and 163 followed up after 12 months intervention (PHP group). AWLD and carers of AWLD were employed as research interviewers. AWLD were full research participants. Results: Annual consultation rates in the intervention and control groups at baseline were low (2.3 and 2.6 visits respectively). A slightly greater increase occurred over the year in the intervention group 0.6 ()0.4 to 1.6) visits ⁄ year compared with controls. AWLD in PHP group reported more health problems at follow-up 0.9 (0.0 to 1.8). AWLD liked their PHP (92%) but only 63% AWLD and 55% carers reported PHP usage. Carers had high turnover (34%). Conclusions: No significant outcomes were achieved by the intervention.