The primary purpose of the Grid Workloads Archive is to provide (anonymized) workload traces from grid environments to researchers and to practitioners alike.
Large-scale multi-site infrastructures (Grids) provide the needed computational support for e-Science. The current situation of the research in resource management in grid resembles strikingly that of the parallel production environments, a decade ago: surprisingly little is known about the real behavior of the studied systems, and most research results are based on empirically composed workloads. Following the analogy with the parallel production environments' world, the evolution of research in grids is tightly connected to the existence and the quality of workload traces from real grid environments.
The goal of the Grid Workloads Archive is to provide a virtual meeting place where practitioners and researchers can exchange grid workload traces. To facilitate the exchanges, a standard Grid Workloads Format (GWF) must be provided. Based on GWF, a high-availability, easily accessible, database of existing grid workload traces must be created. For practitioners, tools that convert workload trace formats specific to different platforms into GWF data must be made available. Finally, links with similar efforts from related research communities (e.g., the Parallel Workloads Archive, PWA) must be established.
Our approach to building the Grid Workloads Archive is:
How can I contribute the the Grid Workloads Archive?
If you have traces from your grid system, either complete or partial, please contact us at gwa@tudelft.nl. We will help you gather the traces, we will anonimize and process the traces, we will publish with your consent the traces, and we will give credit where its due.
Team members (alphabetical order):
Side content