Assessment of Resource Quality for Service Level Agreements in Life Science Grids

Abstract

This article focuses on measuring, describing, monitoring and publishing the quality and performance of grid resources. Life science communities can employ Service Level Agreements (SLAs) with their resource providers to ensure the delivery of services. For this, it is important for both the life science communities and their providers to understand and quantify the performance and service quality of different grid environments. However, measuring service quality in grid infrastructures utilizing different middlewares, as in the German Grid Initiative, is a complex problem. We describe the state of quality metrics which are currently used by the German life science communities MediGRID, Services@MediGRID and PneumoGrid. We also identify further quality metrics for defining and monitoring grid resource quality in D-Grid. It is important to publish and exchange the quality information by grid information systems, which are the entry points to grid services. Therefore, we also present how quality information can be handled by the GLUE v2.0 Schema, which is the upcoming standard data model used by grid information systems. For measuring and monitoring the quality metrics in multi-middleware environments two approaches are discussed. The first approach extracts quality information from an external benchmarking system and loads it to the grid information systems. The second solution targets life science communities that do not utilize legacy benchmarking systems, but operate traditional monitoring systems, like Nagios.