The first objective of this website is to index public workload datasets useful to researchers in the scheduling community and allow quick filters to search through this database. For instance, "what are the available datasets spanning more than a year and reporting the memory used by each job?". This allows to easily find all datasets with the properties relevant to a given experiment. Further objectives include statistical description and analyses of these datasets and a focus on energy requirements, e.g., how to obtain relevant data regarding realistic energy prices, energy mix, renewable source performance, weather data...

We complement the trace datasets with data describing hardware evolutions of the systems on which the dataset was scheduled if available, and trace generator if synthetic data are required. The scope of the datasets indexed is restricted to files describing individually the jobs submitted to the system. For instance, we do not include datasets composed of subsequent descriptions of the total load of a server, nor datasets focused on "Function As A Service" systems where the same functions are called many times and the description of their executions are aggregated over subsequent time periods.

This website has been built by the members of the SCALE working group. You can contribute through this gitlab repository. Each dataset is described by a simple TOML file.